Description: FIELD OF THE INVENTION
[0001] The invention relates a crime analytics platform for law enforcement. The crime analytics platform uses machine learning (ML) and computer vision algorithms and/or techniques to identify crime and violent activities in a geographical area. This analytical platform enables the decision makers such as police forces, special task forces, and other investigation agencies to conduct an in-depth analysis of factors related to criminal activities, determining criminal hotspots, creating criminal profiles, and learning criminal trends and to take preventive measures against occurrence of crimes.
BACKGROUND OF THE INVENTION
[0002] Crime is behavior, either by act or omission, defined by statutory law or common law as deserving of punishment for the wrong doing. Crimes may range from petty crime, such as a thief stealing a purse, to serious crimes, including assault, robbery, sexual violence and murder. The impact of crime in a society makes it difficult for a common man to survive or live in any geographical area. These geographical areas are marked as unsafe areas for living and hinders the economic development. If crime levels rise, there will be less money for other services such as education and healthcare. Various factors such as economic conditions, including median income, poverty level, and job availability, cultural factors and educational, recreational, and religious characteristics, family conditions with respect to divorce and family cohesiveness, affect the occurrence of a crime in the environment. Although most crimes require the element of intent, certain minor crimes may be committed based on strict liability even if the defendant had no specific mindset with regard to the criminal action.
[0003] Geospatially integrated Machine Learning (ML) models and tools are used for crime intelligence and taking effective decisions to tackle crime. Furthermore, new technologies and strategies using Artificial Intelligence (AI) techniques are continuously being implemented to make society free of crimes.
[0004] Researchers, bureaucrats and decision-making bodies are using Artificial Intelligence (AI) techniques and tools to estimate the critical factors/features behind crime patterns, types of crimes happening in the area, underperforming areas in the eyes of law, hotspots of different crimes and visually identifying the potential escape routes of the criminals. Geo-AI tools help in ensuring public safety and increased law and order by reducing the time to identify the suspects. In view of ever
[0005] As per National Crime Records Bureau (NCRB), 66,01,285 cognizable crimes were registered in 2020 in India. There was an increase of 14,45,127 (28.0%) in registration of cases over 2019, when the total number of registered cases were 51,56,158. Crime rate registered per lakh population has increased from 385.5 in 2019 to 487.8 in 2020. Serial crimes usually have a pattern and this pattern is identified to crack and predict the crime. People engaged in identification of patterns of crime need to do this exercise manually, which is tedious, time consuming and sometimes inaccurate. Certain locations are more prone to crime (hotspots). Identifying such locations/hotspots is a challenge and AI techniques and analytical platform can help the police forces to predict possible places accurately. This can help in preventing such crimes and reducing the crime rate.

PROBLEM STATEMENT
[0006] Using predictive analytics, AI systems, besides scanning through volumes of information, can provide accurate predictions on the occurrence of crime. Despite of advanced machine learning models there are a few challenges faced by the stakeholders, for example: the hotspots or most crime prone areas may be identified but the group/set of people who commit crime and the intent may still vary and may not be detected. Hence, there is a need for a crime analytical platform that includes analytics, geospatial data, artificial algorithms and statistical algorithms and other modules to help visualize and statistically analyze certain external and internal factors including demographics associated to different types and volumes of crimes and features that contribute to the crime environment and provokes crime in a society. This platform will help in improving the law and order by giving crime related intelligence to the decision-makers and police forces to take timely decisions to develop suitable strategies and prevent crime beforehand by making data-driven decisions.

SOLUTION
[0007] The invention discloses an improved set of layered solutions which help in estimating crime hotspots along with correlation of the occurrence of a crime and it’s type with the geography and demography of a region. This provides early insights for allocating relevant resources and prevention of crime.

SUMMARY OF THE INVENTION
[0008] A crime analytical platform is disclosed. The crime analytical platform includes an analytics, a geospatial data, an artificial algorithms and a statistical algorithms and other modules to help visualize and statistically analyze certain external and internal factors including demographics associated to different types and volumes of crimes and features for assessment and prevention of crimes.
[0009] The crime analytical platform includes an array of spatial data and advanced machine learning and artificial intelligence techniques in data pre-processing, extracting valuable information, assessing the extracted information and deploying it for optimization of resources. The data processing and machine learning is integrated with artificial intelligence algorithms to analyze different features related to occurrence of crime. These features can then be utilized for improving law and order and for making rational decisions against prevention of crime.
[00010] In some embodiments, the crime analytical platform may identify the causes of criminal behavior, which lie in the social processes and the social structure. For example, the people in society may commit crimes in the process of socialization and due to the emerging opportunities. Due to emerging opportunities, people in society may overlook the difference between the right or the wrong and may be overwhelmed with the enlarging desires, which may act as strong motivation for taking to crime.
[00011] In some embodiments, the crime analytical platform implements a machine learning regression technique, which identifies the hotspots, spatially, seasonally and temporally (shift of hotspots on a timescale) of the crime incidences and links the occurrence of the crime to various factors such as presence of certain geographical features, weather condition, surrounding demographics, living conditions, crimes registered, temporary and permanent police check posts and patrolling, safety facilities and infrastructure, accessibility of the crime locations, escape routes and traffic conditions near surrounding the crime locations, budget disbursed for the training of the officials.
[00012] In some embodiments of the present invention, the crime analytical platform may use crime data, collected from records, surveys and other sources and by assimilation techniques.
[00013] In some embodiments of the present invention, the crime analytics platform may collate the data one or more sources to understand the changing hotspots and patterns of occurrence of crime spatially and temporally, predicts future crimes and present recommendations to avoid crime.
[00014] In some embodiments, the crime analytical platform may include other modules such as geospatial artificial intelligence module, rule based module, data aggregation module apart from other modules. The geospatial artificial intelligence module may include a risk terrain modeling for identifying high, medium and low risk areas. For example, a near repeat analysis may be performed keeping time as constant. In another example, an area suffered from a criminal activity once is likely to be suffered with a criminal activity again with the same or different type of crime. Likewise, an environmental time related analysis may be performed, where in the immediate surrounding factors are identified that may provoke criminal environment. In embodiments, the environmental time related analysis may be performed for hotspot identification and feature engineering.
[00015] In embodiments, the risk terrain modeling may combine data from one or more sources and link them to the geographic features. The crime analytical platform may provide a ‘risk map’, which indicates the areas with high risk of crimes and/or low risk of crimes in a specific region or a specific area, for example, a city. The risk map may be generated using all the geographical features of a specific area, which attracts the crowd for daily/routinely activities like ATMs, bars, gyms, parks, banks or some other type of public utilities. The process of generating the risk map is known as risk terrain modeling.
[00016] In embodiments, the crime analytical platform may provide information on the induction of certain crimes based on the environment, for example, construction sites, dark alleyways, malfunctioned traffic lights, war zones or the like. This analysis is referred in this document as environmental time related analysis. The environmental time related analysis may help the police to identify environment that may lead to crime owing to the nature of the place.
[00017] In some embodiments, the artificial intelligence techniques and machine learning techniques may be applied to predict, analyze and prevent crimes. The machine learning technique implemented in the machine learning modules may include a processor and a memory, which classifies different areas into a high risk, a medium risk or a low risk zone based on various physical and geographical characteristics of an area or a region.
[00018] In some embodiments, geospatial data may be in form of a map, which is used to solve the problems of routes, time, distance and direction.
[00019] In some embodiments, a geospatial intelligence module may be implemented in the crime analytical platform to forecast and solve the problem of prevention of crime.
[00020] In some embodiments, the statistics techniques and algorithm may be combined with Geographic Information Systems (GIS) to model a risk terrain modeling and environmental time related analysis. Similarly understanding the patterns of the crime, highlights the environmental and geographical features affecting the occurrence of the crime.
[00021] In some embodiments, satellite images may provide surrogate relationships between land use and crime. Satellite images or remote sense data may be analyzed by using statistical techniques such as Spatial Autocorrelation, Chi-Squared Analysis, Hot Spot Analysis (using statistically significant spatial clusters of high values (hot spots) and low values (cold spots)), Nearest Neighbor, Near Repeat Analysis.
[00022] In some embodiments, Chi-squared test may be performed on the features that might affect crime.
[00023] The invention uses a data assimilation technique, broadly refers to the abstraction of the real world at certain places.
[00024] In embodiments, the risk terrain modeling may involve the process of attributing qualities of the real world to micro-level places within a terrain or a region or a specific area of interest, and overlaying multiple terrains together to produce a single composite map. The derived value of each place in the composite map holds the value of the compounded crime risk associated with the terrain, the region or the specific area of interest.
[00025] In embodiments, the crime analytical platform may identify and relate the geographical factors associated with one or more type of criminal events. These related factors are further evaluated and analyzed to identify specific places where illegal behavior and new crime incidents will emerge and/or cluster. The crime analytical platform may suggest deployment of resources, create interventions posts, take corrective measure such as increased patrolling, checking, putting up additional remote monitoring etc. to strategically and tactically prioritize efforts to mitigate crime risks.
[00026] In embodiments, the crime analytical platform may evaluate and analyze one or more spatially placed influencing features, which are empirically tested using different artificial intelligence and machine learning algorithms and the value of the risk increases with the increase in empirical relation between the influencing features and the theoretically-grounded spatial associations with known occurrence of a crime incident. Thereafter, the crime analytical platform may select one or more risk factors are selected to produce a terrain model illustrating risk of crime.
[00027] In embodiments, the crime analytical platform may provide escape routing, which uses traffic analysis, risk areas and possible escape routes used by the criminals. The escape route are marked on the map and are statistically analyzed, which provides a real-time insight to the resources to deploy forces in occurrence of crime.
[00028] In embodiments, the crime analytical platform may provide a single window view of the status of crime occurrence, spatial and temporal changes in the occurrences, in terms of location and type, other relevant geographical features. This information may be correlated with the escape routes in view of real-time traffic. In addition, the high risk areas are identified and the overlap of the crime historical hotspots are merged with the risk areas to validate the terrain model for analysis, tracking, forecasting, resource allocation, mitigate crime risk and prevention of crime.

BRIEF DESCRIPTION OF FIGURES

[00029] Fig. 1 illustrates the environment of a computer implemented crime analytical platform in an embodiment of the present invention;
[00030] Fig. 2 illustrates the different components of a computer implemented crime analytical platform in an embodiment of the present invention;
[00031] Fig. 3 illustrates different components of a crime analytics module in an embodiment of the present invention;
[00032] Fig. 4 illustrates different modules of a data aggregation module in an embodiment of the present invention;
[00033] Fig. 5 illustrates a different variables associated with prevention of crime in an embodiment of the present invention;
[00034] Fig. 6 illustrates different component of a statistical analysis module in an embodiment of the present invention;
[00035] Fig. 7 Fig. 7 illustrates a risk heat map indicating high/medium/low risk areas of crime based on the analysis of the crime analytical platform in an embodiment of the present invention;
[00036] Fig 8 shows geographical points of interest along with the risk factor associated with occurrence of crime in an embodiment of the present invention;
[00037] Fig. 9 illustrates demographic variables for a geographical area in an embodiment of present invention;
[00038] Fig. 10 illustrates a table of standard norms for presence of a police station in a geographical area in an embodiment of the present invention;
[00039] Fig. 11 illustrates the geospatial spread of crime and police stations in different geographical areas in an embodiment of the present invention;
[00040] Fig. 12 shows the distribution of types of crimes located in a geographical area in an embodiment of the present invention;
[00041] Fig. 13 shows the distribution of crimes in a geographical area (ward) of a district in an embodiment of the present invention.;
[00042] Fig. 14 illustrates a statistical analysis module in an embodiment of the present invention;
[00043] Fig. 15 illustrates different components of a feature engineering module in an embodiment of the present invention;
[00044] Fig. 16 illustrates different components of risk terrain modeling module in an embodiment of the present invention
[00045] Fig. 15B illustrates different components of environmental time related analysis module in an embodiment of the present invention;
[00046] Fig. 15C illustrates different components of crime hotspot analysis module in an embodiment of the present invention;
[00047] Fig. 15D illustrates different components of traffic and exit route analysis module in an embodiment of the present invention;
[00048] Fig. 15E illustrates different components of police station accessibility analysis module in an embodiment of the present invention;
[00049] Fig. 16 illustrates a geospatial analysis module in an embodiment of the present invention;
[00050] Fig 17 illustrates a process of risk terrain modeling in an embodiment of the invention;
[00051] Fig 18 illustrates a process of environmental time modeling in an embodiment of the invention;
[00052] Fig 19 illustrates a process of hotspot identification in an embodiment of the invention;
[00053] Fig 20 illustrates a process of traffic and exit route analysis in an embodiment of the invention;
[00054] Fig 21 illustrates a process of determining police station accessibility in an embodiment of the invention;
[00055] Fig. 22 illustrates a process of providing recommendation for deployment of resources in an embodiment of the present invention;
[00056] Fig. 23 illustrates a process of data aggregation and data visualisation in an embodiment of the present invention; and
[00057] Fig. 24 illustrates a block diagram for prediction and prevention of crime with missing data in an embodiment with of the present invention.

DETAILED DESCRIPTION OF THE INVENTION
[00058] As used herein any reference to “one embodiment” or “an embodiment” means that a particular element, feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Likewise, the term “in some embodiments” a particular element, feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments. The appearances of the phrase “in one embodiment” or “in one implementation” or “in variation of the implementation” at various places in the specification are not necessarily all referring to the same embodiment.
[00059] As used herein, the terms “comprises,” “comprising,” “includes,” “including,” “has,” “having” or any other variation thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. The use of the “a” or “an” are employed to describe elements and components of the embodiments herein. This is done merely for convenience and to give a general sense of the invention. This description should be read to include one or at least one and the singular also includes the plural unless it is obvious that it is meant otherwise.
[00060] Fig. 1 illustrates an operating environment of a computer implemented crime analytical platform in an embodiment of the present invention. The computer implemented analytical platform environment 100 includes a crime analytical platform 110 connected with one or more geographical areas such a geographical area 102 and a geographical area 104. Each geographical area comprises one or more crime locations for example, a crime location 120A in the geographical area 102 or a crime location 102B in the geographical area 104. In addition, each geographical area may include one or more police stations such as a police station 130A in the geographical area 102 or a police station 130B in the geographical area 104. The crime locations such as the crime location 120A and the crime location 120B may be collectively referred to as crime locations 120 and the police stations may be collectively referred to as police stations 130. In some embodiments, the geographical location 102 may include one or more crime locations and one or more police stations. For example, the crime location 120A falls under the jurisdiction of the police station 130A in the geographical area 102.
[00061] The crime analytical platform 110 is connected to a server 112, a database 114, and a cloud computing environment 118 and other electronic data processing devices through a network 108. The crime analytical platform 110 collects and aggregates data related to one or more crime locations 120 for one or more of the geographical areas such as the geographical area 102 or the geographical area 104 to perform analytics for improving the law and order.
[00062] In some embodiments, the computer implemented crime analytical platform 110 may reside in the server 112.
[00063] In another embodiment, the analytical platform 110 may be implemented over a distributed environment or on a cloud computing environment 118.
[00064] In some embodiments, the crime analytical platform 110 may be connected with one or more databases 114. The one or more databases 114 may include an ancillary database, which may store information related specific crime or profile of different stakeholders associated with the specific crime. For example, the ancillary database may store information regarding the public transportations, road network of the geographical area under consideration, weather conditions, street lighting, and demographics. In addition, the ancillary database in at least one implementation may include information related to geospatial data of the crime, the fund received by the criminals, the crime indicators and other relative factors associated with the crime.
[00065] Fig. 1 illustrates an exemplary embodiment of the operating environment 100 of the crime analytical platform 110 but in other implementations the operating environment 100 may include additional or fewer components than shown.
[00066] The computer implemented crime analytical platform 110 may employ and implement statistical technique like the decision tree regressor for evolutionary computation to construct a model and further combine this model with additional ancillary data but not limited to weather conditions, road and traffic conditions, street lighting, points of interest, economic activity in that geographical area.
[00067] Fig. 2 illustrates the different hardware components of a computer implemented crime analytical platform in an embodiment of the present invention. The crime analytical platform 110 includes a memory 204, a one or more processor 218, an input/output module 220, a communication module 222, an internal bus 214 and an external interface 224. The internal bus 214 allows exchange of data and electrical power between the memory 204 and the processor 218, the input/output module 214 and the communication module 222. Additionally, the external interface 224 allows communication between external devices such as the server 112 or data received from database(s) 114 about crime. In addition, the one or more database(s) 114 may provide information about crime location 120 from one or more geographical areas. The memory 204 may include an operating system 208, a one or more applications 210, and a crime analytics module 212 apart from other modules. The operating system 208 may be a windows OS, Macintosh OS, Linux OS or some other type of operating system. The one or more applications 210 may be related to crime data collection, victim data collection and analysis, crime analytics and other data related to the crime analysis and management. The crime analytics module 212 may include machine learning algorithms, database, and other forecasting algorithms for hotspot analysis, route optimization, and resource management.
[00068] Fig. 2 illustrates an exemplary hardware configuration of the crime analytical platform 110, however, in other implementations, the crime analytical platform 110 may have additional or lesser number of modules.
[00069] Fig. 3 illustrates different components of a crime analytics module in an embodiment of the present invention. The crime analytics module 212 may include a data aggregation module 302, a statistical analysis module 304, a feature engineering module 306, a data integration module 308, a classification module 310, A geospatial analysis module 312, a crime analytics engine 320, and an external database 328 apart from other modules.
[00070] The data collection module 302 may collect data from different geographical areas and regions, for example, the geographical area 102. In addition, the data collection module 302 may also receive data from external sources such as but not limited to external database 328, which includes historical crime data for one or more geographical areas, geographical regions or geographical areas of interest.
[00071] The statistical analysis module 304 may analyse and perform statistical analysis on crime and its related data from different geographical areas. The statistical analysis may provide data patterns for performing analytics related to crime. In some embodiments, the data received form one or more sources may be in different formats. The statistical analysis module 304 may convert the received data from different formats into ASCII format for analysis or superimposing on different maps, which are in ASCII format. In addition, the data received from remote sensing satellite may be received by the data integration module 302 and passed to the statistical analysis module 304. The remote sensing data may provide additional information related to geospatial data such as data related to weather conditions, road conditions, land use conditions and other aspects of the terrain.
[00072] The crime analytics module 212 also includes a feature engineering module 306. The feature engineering module 306 may extract features related to geographical points of interests like, banks, ATMs, restaurants, supermarkets and other aspects to be used for training the decision tree model to perform bivariate classification. The crime analytics module 212 may also perform prediction related the crime (hotspots). The feature engineering module 306 extracts features from one or more data sources, for example, the extracted features may be from a geographical data, a crime data, a demographic data and the like.
[00073] The data integration module 308 may combine features extracted by the feature engineering module 306 and add it to the ASCII data to create meaningful analysis to produce different analytics related to environment, accessibility, location, and weather conditions. The ASCII data is received from the external database 328. In some embodiments, the ASCII data can be received from the geospatial analysis module 312. The ASCII data represents geographical landscapes such as maps, terrains and points of interests and other geospatial indicators such as roads, bridges, rivers etc. In additional, the data integration module 308 may obtain data from other data sources such a remote sensing data. Furthermore, the external database 328 may also provide data related to crime type, crime location, and geographical landscape, which can be used as an input to define different level of granularity. In some embodiments, the data assimilated from different sources and the location intelligence derived from geographical data may be used to train the machine learning algorithms.
[00074] In some embodiments, the data integration module 308 may utilise features extracted by the feature engineering module 306 and combine it with the ASCII data to create meaningful analysis and to produce different analytics related to environment, accessibility, location, and weather conditions.
[00075] The classification module 310 implements statistical analysis algorithms on the data received from a crime analytics engine 320, the data integration module 308 and the geospatial analysis module 312 to produce results that allow a user to draw inferences based on the set goals. In embodiments, the set goals may be related to crime prediction, crime control, crime prevention, identifying crime hot spots or other aspects related to crime. The classification module 310 may also receive geospatial data relate to one or more parameters, which may include data such as includes environment data, accessibility data, location data, and weather conditions. In addition, the classification module 310 may also receive data from the external database 328 such as a crime type, a crime location, and a geographical landscape or other type of data related to crime detection and crime prevention. In one implementation of the embodiment, the set goals are related to crime prevention strategy based on location intelligence, crime data, demographic data and other aspects using machine learning algorithms. The classification module 310 is connected to the crime analytics engine 320. The crime analytics engine 320 includes a rule-based engine 322, a recommendation module 324, an artificial intelligence module 318 and an analytics database 314. The rule-based engine 322 implements different rules for performing crime analytics to provide useful insights for prevention of crime. The analytics database 314 includes data related to crime for different geographical area like the geographical area 102 and contains one or more analytical models for determination of set goals, which were developed during training of artificial intelligence algorithms. The artificial intelligence module 318 can apply one or more analytical models stored in the analytics database 314 and perform the crime related analytics in real time.
[00076] In some embodiments, the classification module 310 and the crime analytics engine 320 may work in tandem to produce crime analytics.
[00077] The geospatial analysis module 312 may receive analyzed data from the classification module 310 and the crime analytics engine 320. The received analyzed data from the machine learning algorithms implemented in the crime analytics engine 320 and the classification algorithms implemented in the classification module 310 is transformed and combined with the geospatial data in raw format such as shape file, tiff, bitmap, jpeg, ASCII or some other type of raster formats or vector formats. The raw format may be analyzed in the geospatial analysis module 312 by different spatial/transformational algorithms to create visualisation data by converting the tiff format back to digital numbers, which provide insights related to crime hotspots linked with other features, for example, demographics, crime locations, type of crime and other information related crime prevention.
[00078] In some embodiments, the geospatial analysis module 312 may also add additional information based on external sources such as but not limited to intelligence received from remote sensing satellite and produce a geo-referenced, projected and classified images related to set goals.
[00079] In some embodiments, the crime analytics engine 320 may be associated with the user interface, which may provide visual and text information related to crime analytics to a user.
[00080] In at least one implementation, the aggregated data from the data aggregation module 302 may form first layer of data and the output after analysis and transformation using statistical analysis module 304 forms a first layered output.
[00081] The analyzed data is passed to feature engineering module 306 for extraction of features related to prediction of set goals from the first layer of data. The extracted feature are displayed as a second layer.
[00082] The extracted features and are passed to the data integration module 308 that combines data from different sources such as data from the external database 328 and the geospatial data from the geospatial analysis module 312 is displayed as a third layer.
[00083] A fourth layer comprises output from the classification module 310 after analysing data from the crime analytics engine 320 combined with the data from the geospatial analysis module 312. Each of the layered data can be displayed independently through a user display or can be provided as aggregated information to achieve the set goals.
[00084] A user can infer and predict results from each layer independently or integrate one or more layer to get insights and perform crime analytics in a customized manner based on the set goals. In embodiments, the layered approach provides insights related to different aspects of crime prevention and crime detection that may be defined in the set goals.
[00085] Fig. 4 illustrates different components of a data aggregation module in an embodiment of the present invention. The data aggregation module 302 includes a proprietary data module 402 comprising proprietary data 402 such as geospatial data, landscape data, navigation data obtained or purchased from third parties; a government data module 404 comprising government data such as crime statistics, crime records, crime hotspots etc.; a crime location data module 408 comprising data collected from one or more crime location and the type of crime in a defined area such as region, state, district or a block and data related to crime in the geographical area. In addition, a demographic data module 410 comprises data collected from different demographic databases. A land use data module 412 comprises data collected from different satellite imageries and town planning agencies show changes in land use and developmental work in different geographical areas. In addition, the data aggregation module 302 further includes research data module 414 comprising of research data from different research institutions related to improvement of quality of law and order, crimes, ethnicity, religiously sensitive area etc. A stakeholder data module 418 comprises data about different stakeholders, such as police station, crime records, fingerprint data, genetic data of criminal records. An ancillary data module 420 comprises data related to public transportation near the school, road network and conditions of the geographical area under consideration, weather data in the geographical region at the time of the crime and CCTV camera locations and dark areas where crime can occur and can go unnoticed.
[00086] In one embodiments, at least one or more data, which is aggregated data by the data aggregation module 302 may be provided by the external database 328. Furthermore, the crime analytical platform 110 may collect data from other sources such as proprietary and non-proprietary databases available by local municipal agencies or other government record keeping agencies.
[00087] Fig. 5 illustrates a different variables associated with prevention of crime in an embodiment of the present invention. The variables provide basis to visualise and understand the various factor related to occurrence of crime in the geographical location. The invention uses one or more statistical tools and techniques to solve the problem of improvement of quality of law and order and to reduce frequency of crimes to render the geographical location as a ‘safe place/area’. The variables are grouped in different categories, each category representing a group of related variables. Examples of category of variables grouped together under different categories include demographics, traffic and exit route, risk terrain modeling (RTM) crime hotspots, police station accessibility, environmental time related analysis and gap analysis as shown in Fig. 5.
[00088] The different groups of variables comprising of one or more variables can be used for building statistical models, machine learning models, and classification models for prevention of crime and in improvement of law and order. The group variables include demographic variables, traffic and exit route variables, risk terrain modeling variables, crime hotspot variables, police station accessibility variables, environmental time related variables and gap analysis variables. Each of the variable group may comprise one or more variables, for example, the demographic variables has 36 variables such as population density, land use, female population etc. The traffic and exit route variable group may include 9 variables such as road intersection density, drive time from crime etc. Likewise, the risk terrain modeling variable group may include 34 variables; the crime hotspot variable group may include 20 variables; the police station accessibility group variable may include 10 variables; the environmental time related group variable group may include 10 variables and the gap analysis group variable include 7 variables. The crime analytical platform 110 may utilize the one or more variables for analysis of terrain and to perform Risk Terrain Modeling (RTM). The RTM select the relevant variables, which can be utilized for statistical model building.
[00089] In one embodiment, the crime analytical platform 110 may use the one or more of the 34 variables for analysis of crime prediction. In another embodiment, the crime analytical platform 110 select variables from each category based on different algorithms and statistical techniques. The relevant variables are utilized for building statistical models and machine learning models. The analytical platform 110 may use different statistical methods/techniques including Spearman Rank-Order Correlation (SROC), a nonparametric test to explore the association strength between variables. For example, different statistical methods can be utilized to determine the associative strength between the social variables and the economic variables. The social variables and the economic variables are drivers of change in the system dynamics of crime and provide information and basis for resource deployment. In other embodiments, the statistical techniques may be applied for data cleansing, removal of correlated data, data outliners and then processed to building machine learning algorithms.
[00090] Fig. 6 illustrates different components of a statistical analysis module for in an embodiment of the present invention. The statistical analysis module 304 comprises a variable analysis module 602 and a statistical data cleaning and variable reduction module 604 apart from other modules. The variable analysis module 602 analyzes the relationship between each dependent and independent variable and determines if it is significant or insignificant with the help of p-test, t-test or chi-squared test. In addition, the variable analysis module 602 also determines if there exists an interrelationship among independent variables or multicollinearity. Multicollinearity refers to the problem when the independent variables are collinear. Variance Inflation Factor (VIF) is an index that provides a measure of how much the variance of an estimated regression coefficient increases due to collinearity. The multicollinearity of the dataset is calculated using VIF.
[00091] The statistical data cleaning and variable reduction module 604 utilises VIF for removing certain variables from the set of independent variables which would result in reduction of variance and provide an optimal set for prediction. A value less than 10 for VIF is aimed for the independent variables. The statistical data cleaning and variable reduction module 604 then removes multicollinearity using a stepwise variable elimination procedure. After removal of multicollinearity, a fixed number of variables, for example 12 variables are selected out of the total number of variables (the total number of variables are 20), which have VIF less than 10. These 12 selected variables were used for model building. A VIF value of ‘infinity’ means that this independent variable has a perfect correlation with other independent variables in the dataset. In embodiments, the statistical data cleaning and variable reduction module 604 may use a Random Forest Decision tree algorithm to calculate the importance scores of variables based on the reduction in the criterion used to select split points.
[00092] Fig. 7 illustrates a risk heat map indicating high/medium/low risk areas of crime based on the analysis of the crime analytical platform in an embodiment of the present invention. The crime analytical platform 110 may initiate the process of building a model by associating the geo-demographic profiling, which links the demographical features surrounding the occurrence of the crime in a specific area with the time, location, and type of crimes. The heat map provides information for resource deployment in different areas for prevention of crime.
[00093] Fig 8 shows geographical points of interest along with the risk factor associated with occurrence of crime in an embodiment of the present invention. In embodiments, the geographical point of interest may be a supermarket, a restaurant, a slum, poverty-sticken areas, a library, a fast food joint, a college, a bus stand, a bar, a bank, an ATM, a school or some other type of geographical point of interest. The significance of each geographical point of interest is determined by calculating the chi-square value and the P-value. Based on the calculated chi-square value and the P-values the significance of each of the geographical point of interest for occurrence of crime is calculated. The law enforcement authorities and the relevant stakeholder can take action and implement intervention strategies for each of the geographical point of interest.
[00094] In some embodiments, the crime and the intensity of crime may be evaluated against the standards set by the government, for example, the number of victims, type of crime, time of crime and other parameters associated with crime.
[00095] In embodiments, the specific geographic area with grave crime type and the dense clustering of crimes are considered to be specific geographic area of high concern.
[00096] The crime analytical platform 110 performs analysis of crime for specific geographic area using chi-squared test for first removing the multicollinearity and redundancy from the dataset, selecting the uncorrelated variables and then applying the exploratory regression analysis, the decision tree regressor and convolutional neural network (CNN) to these selected variables for finding the most important variables affecting the risk maps and determining their significance based on the calculated p-value. The crime analytics engine 320 may implement one or more artificial intelligence models to determine the important factors in order to create risk maps.
[00097] Fig. 9 illustrates demographic variables for a geographical area in an embodiment of present invention. In some embodiments, the specific geographic area 902 may be referred to as ward, a locality, a community area, a municipal area etc. In one exemplary illustration of a ward Fig. 9 lists the demographic variables as shown such as ward population, literacy percentage, male population, main agricultural workers_total apart from other demographics variables. The exemplary illustration provides each demographic variables along with it's corresponding value. For example, in a study area of a ward measuring 1.5 sq. kms; the number of crimes recorded in a geographic area (which is a subset of the study area is 5 units); the number of police stations and police check posts in the geographic area is 1; the number of restaurants in the geographic area are 11 and the total population of the geographic area is 10640.
[00098] After analysing the geo-demographic profiling, the crime analytical platform 110 analyzes the crime information to find gaps in the deployment of resources. The crime analytical platform 110 then compares the actual number of deployed resources to the sanctioned resources and further determines the factors responsible for such a gap in the deployment of resources.
[00099] Fig. 10 illustrates a table of standard norms for presence of a police station in a geographical area in an embodiment of the present invention. As shown, for city 1, the sanctioned police per lakh population is 212.58 whereas, the actual police to population per lakh is 205.76. Likewise for city 2 with a total population of 5,74,840having 43 police stations, the sanctioned police per lakh population is 1224, whereas the actual police to population per lakh ratio of 1184. The government standard for police station accessibility were considered as benchmark standards for performing the gap analysis in a given geographical area. These standards are in form of coverage of population (sanctioned vs. actual) and time taken for the resources to reach the place of crime on time while considering the hours of the day, weather and live traffic.
[000100] The crime analytical platform 110 then performs the next step of assessing the gaps in the escape routes considered by the criminals and the best routes to be considered by the police resources to reach the location on time. The routes are divided into 4 categories – Motorway, Primary, Secondary and Tertiary. In one exemplary analysis, it was found that majority of the crimes occurred on the highways or near it, which facilitates the escape of the criminals. Thus, it is recommended to deploy high alert check posts from the routes leading outside the geographical area/city.
[000101] Fig. 11 illustrates the geospatial spread of crimes and police stations in geographical areas in an embodiment of the present invention. As shown, the different geographical areas (wards) with different demographics may have different crime rate. The crime analytical platform 110 may perform a layered solution to improve the resource deployment in the selected geographical area(s). The crime analytical platform 110 uses buffers and isochrones, which are drawn around each police station to check if the crime location or all corners of the block are accessible from nearest police stations. Furthermore, the crime analytical platform 110 also determines if the drive time from the nearest police station is less than 5 minutes for a geographical area. In embodiments, the drive time may be dependent upon different parameters such as distance from nearest police station, crime rate in that geographical area (ward), points of interest in that geographical area and other variables associated with crime. Based on the crime rate in the geographical area (ward); the buffers and isochrones, which have been drawn around each police station; and the drive time from the nearest police station the resource allocation can be made in the geographical area (ward).
[000102] Referring to Fig. 11, the isochrones and the buffer analysis shows that the police stations covered almost 70% of the total geographical area in a drive time of 5 minutes. Furthermore, the buffers and isochrones show that two biggest blocks in the area were not completely covered. Another layer provides information related to live traffic and actual street network. The live traffic and actual street network is mapped to check the different parts of the roads that are blocked or face heavy jams. In addition, the crime analytical platform 110 determines the time of traffic jams (during day) and further suggests an alternate route for the resources to reach the destination in shortest time. Another layer provides information related to police stations in an around the vicinity of the crime location. Based on the information related to the crime location, the crime analytical platform 110 may alert safety in the adjoining geographical areas. In some embodiments, the crime analytical platform 110 may alert important police stations in different geographical areas of the crime and its location immediately.
[000103] The crime analytical platform 110 may perform computation and prediction of the exit route. In preferred embodiment, this analysis is performed by the analytical engine 320. The artificial intelligence module 318 implements artificial intelligence algorithm/machine learning algorithms to predict exit routes of the criminal and alerting the highways leading outside the city.
[000104] Fig. 12 lists the different types of crime considered by the crime analytical platform in an embodiment of the present invention. The different type of crimes associated with city 2 are related to drugs, fraud, molestation, murder, theft, rape or sexual abuse.
[000105] Fig. 13 shows the distribution of crimes in each block of a district in an embodiment of the present invention. As shown, in an exemplary implementation, the block GA1 in city 2 has highest number of recorded crimes. In embodiments, the crime analytical platform 110 may consider six categories of crimes for analysis and the collective numbers of crimes occurring in an area are observed. In embodiments, one or more variables may be used predicting the occurrence of crimes leading to improvement in quality of law and order.
[000106] Fig 14 illustrates a statistical analysis module in an embodiment of the present invention. The statistical analysis module 304 comprises of an exploratory regression analysis module 1402, a correlation analysis module 1404, a univariate analysis module 1408, a bivariate analysis module 1410, a multivariate analysis module 1412, and a cluster analysis module 1418 apart from other modules. The exploratory regression analysis module 1402 is used to create and analyse the different features using regression analysis and to evaluate the relative importance of each of the features. The correlation analysis module 1404 analyses different type of correlations such positive and negative correlation, linear and non-linear correlation, simple, multiple, and partial correlation within the data and/or features. The features are evaluated using the univariate analysis 1408, the bivariate analysis module 1410 or the multivariate analysis module 1412 or any combination of the univariate analysis 1408, the bivariate analysis 1410, and the multivariate analysis module 1412. The cluster analysis module 1412 analyses the data clusters during the statistical analysis to explore the relationship between different features.
[000107] In some embodiments, the statistical analysis module 304 implements the clustering and dimension reduction techniques to create graphical displays of high-dimensional data containing one or more features. In some embodiments, the geospatial data and statistically data may be combined to produce a spatial representation and visualisation. This spatial representation and visualisation result is then combined or overlaid with the result received from the machine learning algorithm to produce additional insights for improving law enforcement. In preferred embodiments, the improvement in quality of law enforcement is related to better deployment of resources.
[000108] Fig 15 illustrates a feature engineering module in an embodiment of the present invention. The feature engineering module 310 includes a decision tree regressor module 1502, a data visualization module 1504, a Risk Terrain Modeling (RTM) module 1508, an Environmental Time Related Analysis (ETRA) module 1510, a crime hotspot indicator module 1512, a traffic and exit route analysis module 1514, and a police station accessibility module 1518. The feature engineering module 310 identifies the features that are relevant in analysis of determining the crime hotspots and gap areas in prevention of crime. The decision tree regressor module 502 creates a decision tree for the task of regression, which can be used to predict continuous valued outputs instead of discrete outputs. The output of the decision tree regressor 1502 is provided to the data visualization module 1504, which provides graphical data to analyze and understand the outcomes related to occurrence of crime. In some embodiments, the data visualisation module 1504 may be a dashboard, a graphical user interface or a customizable and interactive GUI. The data visualization module 1504 is connected with the RTM analysis module 1508 that feeds features, which impact the occurrence of crime for graphical presentation of risk areas. The ETRA module 1510 provides the impact of one or more features and further identifies the impact of the feature on the crime hotspots and resource deployment. The crime hotspots indication module 1512 performs analysis of the previous crime locations using machine learning algorithms to forecast the hotspots and occurrence of crimes. The data visualization module 1504 also receives data from the traffic and exit route analysis module 1514. In some embodiments, the occurrence of crime and after crime actions may be assessed based on the traffic and exit routes/escape routes taken by criminals. In some embodiments, the police station accessibility and the gaps in resources are also assessed in a separate layer in the feature engineering module 310.
[000109] In some embodiments, the data visualization module 1504 may combine data from multiple sources such as from RTM analysis module 1508, the ETRA module 1510, the crime hotspots indicator module 1512, traffic and exit route analysis module 1514, and police station accessibility module 1518 to arrive at a layered data visualisation to uncover addition insights for the quality of law and order (layer 1). This is further combined with geospatial data to arrive at another layer of insights (layer 2). This is referred as layered approach that provided better insights about the reason for high occurrence of crimes in wards or alternatively good performing wards. In addition, it provides information related to key indicators that result in good and bad performance of some wards.
[000110] In one implementation of this embodiment, the layered data to uncover insights related to crime occurrence and prevention is be provided. For example, the decision tree regressor module 1502 provides the analysed data to the data visualization module 1504, which provides graphical data to analyze and understand the outcomes related to occurrence of crime (layer 1). Layer 2 data may be provided by the RTM analysis module 1508 that feeds features impacting crime for graphical presentation of risk areas. Layer 3 data may be provided by the ETRA module 1510 provides the impact of one or more features and further identifies the impact of the feature on the crime hotspots and resource deployment. Furthermore, another additional layer may be provided by the crime hotspots indication module 1512 performs analysis of the previous crime locations using machine learning algorithms to forecast the hotspots and occurrence of crimes. Likewise, another layer may be provided by the traffic & exit route analysis module 1514. In embodiments, each of the layers may be analyzed separately or superimposed over each other to provide a layer analysis of the crime and deployment of resources.
[000111] In another implementation of the present invention, a layered approach may be implemented by the crime analytical platform 110 for determining the occurrence of crimes and hotspots, for example, the first layer may be associated with analysis of geospatial data.
[000112] A second layer may be associated with determination of accessibility to the police stations.
[000113] A third layer may be associated with identification and selection of dimensions using one or more statistical techniques.
[000114] A fourth layers may be associated with analysis and determination of features impacting occurrence of crimes using machine learning algorithms.
[000115] In different implementations, the different layers can be combined in different ways to provide the set objectives that is, improving the quality of law and order, improving the ward performances in terms of safety or some other objectives.
[000116] Fig. 16 illustrates a geospatial analysis module in an embodiment of the present invention. The geospatial analysis module 312 comprises a spatial autocorrelation module 1602 and a buffer analysis module 1604 apart from other modules. The spatial autocorrelation module 1602 is connected to a geo-demographic profiling module 1608 and a crime intervention module 1610. Furthermore, the buffer analysis module 1604 is connected to an accessibility analysis module 1612 and a law enforcement gap analysis module 1614. The spatial autocorrelation module 1602 exchanges data with the geo-demographic profiling module 1608 and the crime clustering module 1610. The spatial autocorrelation module 1602 identifies the patterns within the spatial data and uncovers these patterns in the geospatial data. The uncovered patterns are applied to geo-demographic profiles of occurrence of crime, location of police stations, road network and other relevant features associated with the law and order. The crime intervention module 1610 evaluates how the cluster of different resources may be organised based on the uncovered patterns.
[000117] The buffer analysis module 1604 is connected with the accessibility analysis module 1612, which determines the accessibility factors to identify different escape routes that may be used by criminals. The analysis of different features are used to identify gaps in the law enforcement gap analysis module 1614. The crime intervention module 1610 evaluates how the cluster of different resources may be organised.
[000118] Fig 17 illustrates a process of risk terrain modeling in an embodiment of the invention. The process 1700 is initiated at step 1702 and immediately moves to step 1704. At step 1704, the process 1700 collects data from multiple sources including geographical data for law enforcement and crime control. The geographical features are extracted from geographical data. At step 1708, the process 1700 performs statistical analysis on the selected features using different statistical techniques to identify data characteristics with respect to statistical parameters, for example, assess the risk based on Chi_squared test. At step 1710, the process 1700 overlays geographical feature on the geospatial data to provide data visualization. The feature analysis and processing may involve implementation of the machine learning algorithm to arrive at set objective of law enforcement, for example, crime control, better deployment of resources, crime investigation, and actionable intelligence to investigate crime that has been committed in any geographical area a few hours back. At step 1712, the process 1700 may assign risk value to each area of the geographical features. At step 1714, the process 1700 may identify geospatial points based on specific land use for occurrence of crime. At step 1718, the process 1700 may predict high risk crime locations to provide recommendation for deployment of resources as per set objectives. Finally, the process terminates at step 1720.
[000119] Fig 18 illustrates a process of environmental time modeling in an embodiment of the invention. The process 1800 is initiated at step 1802 and immediately moves to step 1804. At step 1804, the process 1800 determines the environmental factors influencing the crime. In some embodiments, the environmental factors may include demographic factors. In some embodiments, the environmental factors may be analysed using machine learning algorithms. At step 1808, the process 1800 assimilates geospatial data influencing the crime. At step 1810, the process 1800 determines the crime hotspots zone in a specific geographical areas such as district, wards etc. At step 1812, the process integrates the geospatial data of crime hotspot zones with the environmental variables. At step 1814, the process 1800 marks the areas of high/medium/low risk area. In addition, the process 1800 at step 1814 provides marking or data visualization points on the geospatial maps associated with high, medium and low risk crime locations for geographical areas. The process 1800 may further provide allocation of resources and location of check posts based on the step objectives. The process 1800 may terminate at step 1818.
[000120] Fig 19 illustrates a process of hotspot identification in an embodiment of the invention. The process 1900 is initiated at step 1902 and immediately moves to step 1904. At step 1904, the process 1900 may assimilate geospatial data related to historical crime records. At step 1908, the process 1900 may perform demographic variables analysis associated with crime. At step 1910, the process 1900 may select demographic variables to determine feature that contribute to crime. At step 1912, the process 1900 creates a geospatial layer of the selected demographic features. At step 1914, the process 1900 may provide geospatial interaction of feature maps. At step 1918, the process 1900 may generate geospatial crime hotspots locations on the map. Finally, the process 1900 terminates at step 1920.
[000121] Fig 20 illustrates a process of traffic and exit route analysis in an embodiment of the invention. The process 2000 is initiated at step 2002 and immediately moves to step 2004. At step 2004, the process 2000 may create geospatial categorization of routes and roads types. At step 2008, the process 2000 may map crime locations and highway exit points. At step 2010, the process 2000 creates and identifies all possible exit routes from the crime location. At step 2012, the process 2000 creates different time isochrones from police stations to find gaps in the coverage. In embodiments, the different type isochrones may be of different time periods, for example, 2 minutes, 5 minutes, 7 minutes or a number ranging from 1 minute to 20 minutes. At step 2014, the process 2000 may map road intersection density with geospatial data. At step 2018, the process 2000 assimilates information to show real time exit routes and possible escape route locations for action by the police. The process 2000 terminates at step 2020.
[000122] Fig 21 illustrates a process of determining police station accessibility in an embodiment of the invention. The process 2100 is initiated at step 2102 and immediately moves to step 2104. At step 2104, the process 2100 maps regions of the selected geographical area with geospatial data. At step 2108, the process 2100 maps existing police stations in the selected geographical area with the geospatial data. At step 2110, the process creates buffer zones around each police station of specific distances. In one exemplary embodiment, the specific distances may be 500 mts, 1000 mts and 2000 mts from each of the selected police station. At step 2112, the process 2100 extracts population density, historical crime and area covered for each of the buffer zone for analysis. At step 2114, the process 2100 analyzes accessibility issues to identify gaps in the police station coverage area, for example, the police station lying outside the designated buffer zones. At step 2118, the process 2100 maps areas to be covered by the law enforcement in the geospatial landscape. The process 2100 terminates at step 2120.
[000123] Fig. 22 illustrates a process of providing recommendation for deployment of resources in an embodiment of the present invention. The process 2200 starts at step 2202 and immediately moves to step 2204. At step 2204, the process 2200 collects data from multiple sources. In preferred embodiment, the process 2200 receives proprietary data from the proprietary data module 402 and the demographic data from the demographic data module 410 and other data from other modules. In addition, the process 2200 may also receive criminal records data, research data, stakeholder holder data, government data and ancillary data from different modules. In some embodiments, the process 2200 may check whether there is missing data, if so, the process 2200 may request for missing data from the data aggregation module 302. Simultaneously, the process 2200 may send an alert to the crime analytical platform 110 to collect the missing data. The data aggregated at step 2204 is analyzed and processed using one or more type of statistical techniques. At step 2208, the software implemented statistical techniques are utilised for data transformation, filtering and choosing the relevant dimensions. For example, the correlation analysis, determining the variance inflation factor and for dropping variables highly correlated variables. At step 2210, the process 2200 performs feature engineering using one or more statistical techniques. The feature engineering includes analysing the features using decision tree regressor and performing variable weightage calculation to extract dimensions/variables, which can be utilised for prediction of quality of law and order. At step 2212, the process 2200 performs geospatial analysis based on different parameters such as geo-demographic profiling of the crime locations, crime intervention analysis, law enforcement gap analysis and other parameters related to occurrence of crimes. For example, the geo-demographic profiling may use spatial autocorrelation. Likewise, the cluster analysis may use K-means algorithm for analyzing the crime intervention analysis. A buffer analysis may be performed to analyze law enforcement gaps and accessibility of police stations. Once the relevant data for prediction is collected and prepared for analysis then at step 2214, the process 2200 applies machine learning and/or artificial intelligence algorithms to predict the crime occurrences. The predicted results are analyzed and passed to the recommendation module. At step 2218, the process 2200 provides recommendation for predicting and improving the quality of law and order.
[000124] In some embodiments, the data collected at step 2204 may be processed in parallel to produce data analytics such as producing performance grading indicators and spatial representation and visualization to improve the resource deployment.
[000125] In some other embodiments, the data collected at step 2204 may be processed in parallel in different layers improve the resource deployment and police accessibility.
[000126] The process 2200 terminates at step 2220.
[000127] Fig. 23 illustrates a process of data aggregation and data visualisation in an embodiment of the present invention. The process 2300 starts at step 2302 and immediately moves to step 2304. At step 2304, the process 2300 collects data from different sources. At step 2308, the process 2300 determines if the there is any missing data. If there is missing data then at step 2310, the process 2300 sends an alert for missing data to the data aggregation module 302. If there is no missing data the process 2300 moves to step 2318 to apply machine learning/artificial intelligence algorithms to predict the occurrence of crime. The data collected at step 2304 may be passed for analyzes and identification of key performance grading indicators. For example, the data collected at step 2304 may combine propriety data with data from other modules and conclude that a key indicator is accessibility of a geographical area from a police station. Additionally, at step 2314, the process 2300 may perform spatial representation on the aggregated and infer that a key indicator in improvement in law and order is resource deployment in high-risk areas. The results of key indicators and the spatial representation may be combined with the prediction received from the machine learning algorithms to provide new insight for improvement of quality of law and order.
[000128] At step 2320, the process 2300 may combine the outcomes from the machine learning/artificial intelligence module for improvement of quality of law and order and the data analytics of steps 2312 and 2314 to provide recommendations and insights for deployment of resources. The process 2300 may terminate at step 2322.
[000129] Fig. 24 illustrates a block diagram for prediction and prevention of crime with missing data in an embodiment with of the present invention. The crime analytical platform 110 may receive data from different sources such as the propriety data module 304, the demographic data module 410, the crime data module 404, the police station data module 408, the criminal dataset module 412, the stakeholder data module 418 and the ancillary data module 420. The data received from different sources are evaluated for any missing data that cause incorrect or false prediction. A missing data module 2402 evaluates if the missing data can cause wrong prediction for crime prevention. If the missing data module 2402 finds that there is missing data, which may affect the prediction of law enforcement then the alert module for missing data 2404 sends an alert to the crime analytical platform 110 for taking correcting action. In one embodiment, the missing data module 2402 and the alert module for missing data 2404 may be the part of the data aggregation module 302.
[000130] If the missing data module 2402 find there is no missing data then it passes the data to analyse to the statistical analysis module 304. The statistical analysis module 304 passes to data to the feature engineering module 306. The feature engineering module 306 passes the data to the geospatial analysis module 312 and finally to the artificial intelligence module 318. The missing data module 2402 may also provide data to a crime prediction assessment module 2408, which may after assessment of the prediction data pass it to a spatial representation and visualization module 2410. The analysed data from the representation and visualization module 2410 and the result received from the artificial intelligence module 318 is passed to the result analysis and recommendation module 2412.
[000131] In embodiments the process 2400 may be implemented as a layered solution with the collected data. At the first layer the process 2400 analysis the data by performing statistical analysis 2208, feature engineering 2210, geospatial analysis 2212, and then using machine learning algorithms to produce results. In parallel, a second layer analysis the aggregated data by performing crime prediction assessment and the spatial analysis/visualization. The results of layer one and layer two are combined and processed using at least one of a statistical technique and/or machine learning algorithms to arrive at in depth insight on factor for improvement of hotspot zones and risk areas for better law enforcement.
[000132] In another embodiment, a three-layered approach may be implemented by using the first layer and the second layer as described. In addition, a third layer for accessibility analysis may be appended to arrive at recommendations. In yet another embodiment, another additional layer of key performance indicators may be added to arrive at recommendations.
[000133] The invention can be modified into many variations in different implementations and is not limited to different embodiments described herein. Other variations that can be amended or modified are within the scope of this invention.
, Claims:WE CLAIM:
1. A crime analytical platform (110) for predicting and controlling the crime in one or more geographical area (102, 104), the crime analytical platform (110) comprising:
a memory (204) comprising a crime analytics module (212), the crime analytics module (212) comprising:
a data aggregation module (302) for aggregating data related to crime from one or more crime locations (120) in one or more geographical areas;
a statistical analysis module (304) configured to analyze and perform statistical analysis on crime data collected by the data aggregation module (320) from one or more geographical areas;
a feature engineering module (306) extract features from data related to crime for one or more geographical areas;
a data integration module (308) to combines features extracted from the feature engineering module (306) with the geospatial data received from a geospatial analysis module (312) and an external database (328), wherein the geospatial analysis module (312) collected geospatial data related to one or more geospatial parameters and the external database (328) provides other data;
a crime analytics engine (320) comprising a rule-based engine (322), a recommendation module (324), an artificial intelligence module (318) and an analytics database (314), wherein
the rule-based engine (322) implements different rules related to performing crime analytics,
the analytics database 314 stores one or more analytics model for crime prevention,
the artificial intelligence module (318) applies analytical models to perform the crime related analytics in real time,
the recommendation module (324) for prediction of crime related to one or more geographical areas;
a classification module (310), which implement statistical analysis algorithms and combines recommendations from the crime analytics engine (320) with the geospatial data and the other data in a layered approach to produce results, which allow a user to draw inferences based on the set goals.

2. The one or more geospatial parameters as claimed in claim 1, further includes environment data, accessibility data, location data, and weather conditions

3. The other data as claimed in claim 1, further includes at least one of a crime type, a crime location, and a geographical landscape.

4. The set goals as claimed in claim 1 are related to prediction of crime in one or more geographical area.

5. The aggregated data collected by the data integration module (308) as claimed is claim 1, includes data related to point of interests.

6. A computer implemented method for prediction and prevention of crime using deep learning algorithms using a crime analytics module (212) stored in the memory (204) having encoded instructions, when executed by a processor (218) cause the computer implemented method to perform steps of:
aggregating data related to crime from one or more crime locations (120) in one or more geographical areas;
performing statistical analysis on the aggregated data collected from one or more geographical areas;
extracting features from data related to crime for one or more geographical areas;
combining extracted features with a geospatial data related to one or more geospatial parameters and other data, wherein the geospatial data include data in ASCII format related to geographical landscape, and
classifying the combined geospatial data and the other data other data to produce results that allow a user to draw inferences based on the set goals.

7. The one or more geospatial parameters as claimed in claim 1, further includes environment data, accessibility data, location data, and weather conditions

8. The other data as claimed in claim 1, further includes at least one of a crime type, a crime location, and a geographical landscape.

9. The set goals as claimed in claim 1 are related to prediction of crime in one or more geographical area.

10. The aggregated data collected by the data integration module (308) as claimed is claim 1, includes data related to point of interests.