Description:FORM 2

THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENT RULES, 2003

COMPLETE SPECIFICATION
(See Section 10 and Rule 13)

Title of invention:
SYSTEMS AND METHODS FOR PREDICTING KEY PERFORMANCE INDICATORS FOR ENTITIES USING DATA DRIVEN BALANCING TECHNIQUES

Applicant:
Tata Consultancy Services Limited
A company Incorporated in India under the Companies Act, 1956
Having address:
Nirmal Building, 9th Floor,
Nariman Point, Mumbai 400021,
Maharashtra, India

The following specification particularly describes the invention and the manner in which it is to be performed.
TECHNICAL FIELD
[001] The disclosure herein generally relates to key performance indicators (KPIs), and, more particularly, to systems and methods for predicting key performance indicators for entities using data driven balancing techniques.

BACKGROUND
[002] Predicting Key Performance Indicators (KPIs) is a common requirement across organizations and enterprises. Conventionally various methods have been proposed to arrive at reasonably accurate predictions. These existing methods used statistical analysis, machine learning, artificial intelligence, etc. For instance, statistical analysis includes use of only historical data and hence may not be completely accurate and reliable. Some of these used distribution pattern over and above other methods. For instance, some of the methods mentioned earlier work better when granularity of data is low. For highly granular data, accuracy reduces significantly in terms of prediction. Further, despite these efforts, the accuracy is not up to the mark and needs some improvement.

SUMMARY
[003] Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems.
[004] For example, in one aspect, there is provided a processor implemented method for predicting key performance indicators for entities using data driven balancing techniques. The method comprises obtaining an input comprising one or more Key Performance Indicators (KPIs) specific to a first time period, wherein the one or more KPIs are associated with a dataset of interest pertaining to an entity; predicting a distribution of the one or more KPIs in a plurality of time intervals comprised in the first time period based on a period specific forecast to obtain a set of predicted distribution of the one or more KPIs; applying a plurality of data driven balancing techniques on the set of prediction distribution of the one or more KPIs to obtain a set of updated data driven balancing based KPIs, wherein the step of applying a first data driven balancing technique amongst the plurality of data driven balancing techniques comprises performing a comparison of (i) each time slot in a current time interval amongst the plurality of time intervals for which the one or more Key Performance Indicators (KPIs) are to be forecasted and (ii) a corresponding time slot of a previous time interval; and identifying one or more changing parameters for the current time interval based on the comparison; wherein the step of applying a second data driven balancing technique amongst the plurality of data driven balancing techniques comprises analyzing an impact of a prominent time period on the one or more Key Performance Indicators (KPIs); wherein the step of applying a third data driven balancing technique amongst the plurality of data driven balancing techniques comprises analyzing an impact of a prominent time slot on the one or more KPIs based on a type of the prominent time slot, a number of locations, and one or more KPI affecting projects; analyzing the set of updated data driven balancing based KPIs with respect to a historical data to obtain one or more time slots and one or more candidate KPI patterns; applying a machine learning model on at least one time interval being selected based on the one or more time slots and the one or more candidate KPI patterns to obtain a predicted KPI pattern.
[005] In an embodiment, the method further comprises re-adjusting the one or more KPIs based on the predicted KPI pattern to obtain one or more re-adjusted KPIs.
[006] In an embodiment, the impact of the prominent time slot is analysed based on a decision tree comprising one or more impact probabilities.
[007] In an embodiment, for each leaf node in the decision tree one or more rules are configured to adjust a KPI distribution percentage.
[008] In another aspect, there is provided a processor implemented system for predicting key performance indicators for entities using data driven balancing techniques. The system comprises: a memory storing instructions; one or more communication interfaces; and one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to obtain an input comprising one or more Key Performance Indicators (KPIs) specific to a first time period, wherein the one or more KPIs are associated with a dataset of interest pertaining to an entity; predict a distribution of the one or more KPIs in a plurality of time intervals comprised in the first time period based on a period specific forecast to obtain a set of predicted distribution of the one or more KPIs; apply a plurality of data driven balancing techniques on the set of prediction distribution of the one or more KPIs to obtain a set of updated data driven balancing based KPIs, wherein applying a first data driven balancing technique amongst the plurality of data driven balancing techniques comprises performing a comparison of (i) each time slot in a current time interval amongst the plurality of time intervals for which the one or more Key Performance Indicators (KPIs) are to be forecasted and (ii) a corresponding time slot of a previous time interval; and identifying one or more changing parameters for the current time interval based on the comparison; wherein applying a second data driven balancing technique amongst the plurality of data driven balancing techniques comprises analyzing an impact of a prominent time period on the one or more Key Performance Indicators (KPIs); wherein applying a third data driven balancing technique amongst the plurality of data driven balancing techniques comprises analyzing an impact of a prominent time slot on the one or more KPIs based on a type of the prominent time slot, a number of locations, and one or more KPI affecting projects; analyze the set of updated data driven balancing based KPIs with respect to a historical data to obtain one or more time slots and one or more candidate KPI patterns; apply a machine learning model on at least one time interval being selected based on the one or more time slots and the one or more candidate KPI patterns to obtain a predicted KPI pattern.
[009] In an embodiment, the one or more hardware processors are further configured by the instructions to re-adjust the one or more KPIs based on the predicted KPI pattern to obtain one or more re-adjusted KPIs.
[010] In an embodiment, the impact of the prominent time slot is analysed based on a decision tree comprising one or more impact probabilities.
[011] In an embodiment, for each leaf node in the decision tree one or more rules are configured to adjust a KPI distribution percentage.
[012] In yet another aspect, there are provided one or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause predicting key performance indicators for entities using data driven balancing techniques by obtaining an input comprising one or more Key Performance Indicators (KPIs) specific to a first time period, wherein the one or more KPIs are associated with a dataset of interest pertaining to an entity; predicting a distribution of the one or more KPIs in a plurality of time intervals comprised in the first time period based on a period specific forecast to obtain a set of predicted distribution of the one or more KPIs; applying a plurality of data driven balancing techniques on the set of prediction distribution of the one or more KPIs to obtain a set of updated data driven balancing based KPIs, wherein the step of applying a first data driven balancing technique amongst the plurality of data driven balancing techniques comprises performing a comparison of (i) each time slot in a current time interval amongst the plurality of time intervals for which the one or more Key Performance Indicators (KPIs) are to be forecasted and (ii) a corresponding time slot of a previous time interval; and identifying one or more changing parameters for the current time interval based on the comparison; wherein the step of applying a second data driven balancing technique amongst the plurality of data driven balancing techniques comprises analyzing an impact of a prominent time period on the one or more Key Performance Indicators (KPIs); wherein the step of applying a third data driven balancing technique amongst the plurality of data driven balancing techniques comprises analyzing an impact of a prominent time slot on the one or more KPIs based on a type of the prominent time slot, a number of locations, and one or more KPI affecting projects; analyzing the set of updated data driven balancing based KPIs with respect to a historical data to obtain one or more time slots and one or more candidate KPI patterns; applying a machine learning model on at least one time interval being selected based on the one or more time slots and the one or more candidate KPI patterns to obtain a predicted KPI pattern.
[013] In an embodiment, the one or more instructions which when executed by the one or more hardware processors further cause re-adjusting the one or more KPIs based on the predicted KPI pattern to obtain one or more re-adjusted KPIs.
[014] In an embodiment, the impact of the prominent time slot is analysed based on a decision tree comprising one or more impact probabilities.
[015] In an embodiment, for each leaf node in the decision tree one or more rules are configured to adjust a KPI distribution percentage.
[016] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS
[017] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles:
[018] FIG. 1 depicts an exemplary system for predicting key performance indicators for entities using data driven balancing techniques, in accordance with an embodiment of the present disclosure.
[019] FIG. 2 depicts an exemplary flow chart illustrating a method for predicting key performance indicators for entities using data driven balancing techniques implemented by the system of FIG. 1, in accordance with an embodiment of the present disclosure.
[020] FIG. 3 depicts a portion of a decision tree with a condition leaf node illustrating various conditions and associated sub conditions as implemented by the system of FIG. 1, in accordance with an embodiment of the present disclosure.
[021] FIGS. 4A through 4C, with reference to FIGS. 1 through 3, depict the decision tree as implemented by the system 100 of FIG. 1, in accordance with an embodiment of the present disclosure.

DETAILED DESCRIPTION OF EMBODIMENTS
[022] Exemplary embodiments are described with reference to the accompanying drawings. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the scope of the disclosed embodiments.
[023] Forecasting/Prediction of Key Performance Indicators (KPIs) (e.g., billing, potential/candidate likelihood of sales prediction, future candidate customer engagements, cost projections at various levels (e.g., day level, monthly level, quarterly, yearly, etc. for an entity and/or for various departments/business units/groups within the entity), and the like.) is a common requirement across organizations and enterprises. Various methods have been proposed to arrive at reasonably accurate forecasts. These existing methods used statistical analysis, machine learning, artificial intelligence, etc. Some of these used distribution patterns over and above other methods. Despite these efforts, the accuracy is not up to the mark and needs some improvement. As explained above, accuracy of forecast is very important for prediction of various KPIs. The expression ‘prediction’ and ‘forecasting’ may be interchangeably used herein. However, many automated as well as manual methods do not provide the desired accuracy. Some of the existing methods that forecast still face certain challenge such as: statistical methods are preliminary, deep learning method may or may not provide accuracy as compared to machine learning (ML) methods, based on the unknown distribution of data. Further, these conventional methods work better when granularity is low (e.g., quarterly, monthly, etc.). For highly granular data (e.g., daily, hourly, etc.) the forecasting accuracy reduces significantly. Such challenges may result in less than expected forecast/prediction accuracy, thereby leading to reduced confidence in various stakeholders in an entity (e.g., organization). Conventionally available machine learning model(s) (e.g., AutoML tools) provide the configuration screens for known methods of accuracy improvement, however these are applicable to limited and specific use cases.
[024] Further, weekends are different in different months, and the number of days for every month are not same. Hence, leap year also creates challenges. Therefore, mapping the previous period with forecasted month is a unique combination every time. All in all, adjusting weekends to improve forecast accuracy, at scale, is a huge task. If this complexity is true for just one KPI, one can only imagine the case of integrated planning scenario in the organization where multiple KPIs are needed to be planned simultaneously, with their own hierarchies. Thus, weekend adjustment becomes a humongous task. Current methods and technologies do not infuse (business) intelligence in each and every use case. One of the KPI predictions, for example billing is considered. Billing date for same customer every month, cannot be same. New customers are added every month, their billing cycle is new. Billing vitals are reviewed for a few projects. These and other PESTLE factors across various geographies lead to fluctuations in billing prediction. As billing is calculated for every day, every week, and every month, it has various challenges to identify the right factor that influences the billing. Regular machine learning solution derived using only historical data of billing, and not considering any other factors, is prone to prediction variance. The weekend adjustments carried out (e.g., refer India application 202221041902, filed on July 21, 2022) are confined to general weekends only. For other influencing factors, new methods are needed.
[025] Embodiments of the present disclosure provide systems and methods that implement different data factors, such as special weekends, holidays, KPI projections, PESTLE (Political, Economic, Social, Technological, Legal and Environmental) factors, etc. to arrive at business rules/technique(s) that improves accuracy, derive data with similar patterns from existing data, and then use machine learning algorithms to arrive at a better prediction. The system and method analyze holidays due to festivals, government and public holidays and other types of holidays, in each country, that can potentially impact billing of respective customers from those countries. PESTLE data which states different factors like political, economic, social, technological, legal and environment data which can be collected with respect to billing geographic currency and then mapped with other data factors, which results in building business rules which can be used in prediction of billing. The system and method of the present disclosure observed/determined a strong correlation between billing projection and billing prediction. When incorporated by the system of the present disclosure, the prediction accuracy improved. In addition, the influence of KPI projection on KPI prediction has also been analyzed.
[026] Referring now to the drawings, and more particularly to FIGS. 1 through 3, where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments, and these embodiments are described in the context of the following exemplary system and/or method.
[027] FIG. 1 depicts an exemplary system 100 for predicting key performance indicators for entities using data driven balancing techniques, in accordance with an embodiment of the present disclosure. The system 100 may also be referred to as a prediction system or a forecasting system and may be interchangeably used herein. In an embodiment, the system 100 includes one or more hardware processors 104, communication interface device(s) or input/output (I/O) interface(s) 106 (also referred as interface(s)), and one or more data storage devices or memory 102 operatively coupled to the one or more hardware processors 104. The one or more processors 104 may be one or more software processing components and/or hardware processors. In an embodiment, the hardware processors can be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, and/or any devices that manipulate signals based on operational instructions. Among other capabilities, the processor(s) is/are configured to fetch and execute computer-readable instructions stored in the memory. In an embodiment, the system 100 can be implemented in a variety of computing systems, such as laptop computers, notebooks, hand-held devices (e.g., smartphones, tablet phones, mobile communication devices, and the like), workstations, mainframe computers, servers, a network cloud, and the like.
[028] The I/O interface device(s) 106 can include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like and can facilitate multiple communications within a wide variety of networks N/W and protocol types, including wired networks, for example, LAN, cable, etc., and wireless networks, such as WLAN, cellular, or satellite. In an embodiment, the I/O interface device(s) can include one or more ports for connecting a number of devices to one another or to another server.
[029] The memory 102 may include any computer-readable medium known in the art including, for example, volatile memory, such as static random-access memory (SRAM) and dynamic-random access memory (DRAM), and/or non-volatile memory, such as read only memory (ROM), erasable programmable ROM, flash memories, hard disks, optical disks, and magnetic tapes. In an embodiment, a database 108 is comprised in the memory 102, wherein the database 108 comprises information pertaining to various key performance indicators (e.g., billing, and so on) of one or more entities (e.g., organization, companies, financial institution, business entities, and the like). The database 108 further comprises historical data pertaining to the KPIs, dataset(s), entities, and the like. The memory 102 comprises one or more Artificial Intelligence (AI) or Machine Learning (ML) models as known in the art. The system 100 or the one or more hardware processors 104 invoke the one or more Artificial Intelligence (AI) or Machine Learning (ML) models to perform the method and/or the steps of the method described herein. The memory 102 further comprises (or may further comprise) information pertaining to input(s)/output(s) of each step performed by the systems and methods of the present disclosure. In other words, input(s) fed at each step and output(s) generated at each step are comprised in the memory 102 and can be utilized in further processing and analysis.
[030] FIG. 2, with reference to FIG. 1 depicts an exemplary flow chart illustrating a method for predicting key performance indicators for entities using data driven balancing techniques implemented by the system 100 of FIG. 1, in accordance with an embodiment of the present disclosure. In an embodiment, the system(s) 100 comprises one or more data storage devices or the memory 102 operatively coupled to the one or more hardware processors 104 and is configured to store instructions for execution of steps of the method by the one or more processors 104. The steps of the method of the present disclosure will now be explained with reference to components of the system 100 of FIG. 1, and the flow diagram as depicted in FIG. 2. Although process steps, method steps, techniques or the like may be described in a sequential order, such processes, methods, and techniques may be configured to work in alternate orders. In other words, any sequence or order of steps that may be described does not necessarily indicate a requirement that the steps be performed in that order. The steps of processes described herein may be performed in any order practical. Further, some steps may be performed simultaneously.
[031] At step 202 of the method of the present disclosure, the one or more hardware processors 104 obtain an input comprising one or more Key Performance Indicators (KPIs) specific to a first time period (e.g., say one or more years). The one or more KPIs are associated with a dataset of interest pertaining to an entity. The system and method have considered invoice details or invoice data as the one or more KPIs. It is to be understood by a person having ordinary skill in the art or person skilled in the art that such examples of considering invoice data as KPIs shall not be construed as limiting the scope of the present disclosure. Other KPIs that can be considered as input to the system 100 for obtaining updated forecasted output may include but are not limited to, potential/candidate likelihood of sales prediction, future candidate customer engagements, cost projections at various levels (e.g., day level, monthly level, quarterly, yearly, etc. for an entity and/or for various departments/business units/groups within the entity), and the like. The first time period may refer to a specific year (e.g., say current year), in an example embodiment of the present disclosure. Below Table 1 illustrates invoice details for the specific year.
Table 1
Month Invoice Amount
January 239757923
February 217388534
March 303261002
April 226977613
May 246008848
June 300779675
July 220633933
August 237558115
September 307869206
October 248010120
November 244765847
December 332710947

[032] At step 204 of the method of the present disclosure, the one or more hardware processors 104 predict a distribution of the one or more KPIs in a plurality of time intervals comprised in the first time period based on a period specific forecast to obtain a set of predicted distribution of the one or more KPIs. In an embodiment of the present disclosure, the system 100 predicts/estimates daily distribution of the one or more KPIs for each month comprised in the first time period (e.g., specified year) based on a monthly forecast/monthly prediction to obtain a set of predicted distribution of the one or more KPIs.
[033] The steps 202 and 204 are better understood by way of the following description: Input data in the form of KPIs/invoice data is received from enterprise database(s) (e.g., the database 108). Further, a monthly forecast is estimated from the received input data using one or more machine learning (ML) models as known in the art. To estimate the monthly forecast, first the input data along with key performance indicator is read from database(s). The input data is preprocessed to ensure that it can be fitted into a forecasting model (e.g., a ML model). Further, an appropriate forecasting algorithm (e.g., as known in the art algorithm) is applied on the preprocessed data to determine the best machine learning model. The best machine learning model is trained with the input data at month level to estimate the monthly forecast with a forecasting number. After estimating the monthly forecast, daily distribution of KPI is copied from previous period for all months in a year. Here, the previous period includes the previous year’s same month and previous quarter same month. However, the previous period may vary depending on the application and requirements for one or more case scenarios under consideration. Below Table 2 depicts the predicted distribution of the KPIs by way of illustrative examples:
Table 2
Day of the month Quarterly prediction August 2021 Yearly prediction August 2021
1 0 0
2 0 0
3 1 1
4 1 1
5 1 1
6 3 3
7 0 0
8 0 0
9 2 2
10 2 2
11 1 1
12 2 2
13 3 3
14 0 0
15 0 0
16 3 3
17 2 2
18 3 3
19 3 3
20 3 3
21 1 0
22 1 0
23 4 2
24 3 3
25 5 5
26 7 7
27 11 11
28 7 6
29 4 6
30 18 17
31 13 12

[034] Further, two kinds of predictions are obtained namely: (i) yearly predictions which are obtained by using the billing percentage numbers from same month previous year of the target month, and (ii) quarterly predictions are obtained by using the billing percentage numbers from same month previous quarter of the target month. More specifically, at step 206 of the method of the present disclosure, the one or more hardware processors 104 apply a plurality of data driven balancing techniques on the set of prediction distribution of the one or more KPIs to obtain a set of updated data driven balancing based KPIs. Firstly, a first data driven balancing technique amongst the plurality of data driven balancing techniques is applied on the set of prediction distribution of the one or more KPIs (e.g., refer weekend adjustment described in India application 202221041902, filed on July 21, 2022). This first data driven balancing technique includes comparing (i) each time slot in a current time interval amongst the plurality of time intervals for which the one or more Key Performance Indicators (KPIs) are to be forecasted and (ii) a corresponding time slot of a previous time interval. In other words, the system 100 performs a comparison of each day (e.g., serving as a time slot) in a current month (e.g., current time interval) amongst the plurality of time intervals (of the several months of the first time period – say current year) for which the one or more Key Performance Indicators (KPIs) are to be forecasted and (ii) a corresponding time slot (e.g., corresponding day) of a previous time interval (e.g., say same month previous year or previous quarter month). Based on the comparison, one or more changing parameters for the current time interval are identified.
[035] The system 100 then applies a second data driven balancing technique amongst the plurality of data driven balancing techniques. The second data driven balancing technique includes analysis of an impact (also referred to as impact analysis) of a prominent time period on the one or more Key Performance Indicators (KPIs). In an embodiment, the prominent time period could include, but is not limited to, special weekend – last week of the current month with extended holiday (e.g., Saturday, Sunday, and Monday or Friday, Saturday, and Sunday in the last week of the specific month. In other words, the second data driven balancing technique includes analyzing impact of the special weekend on the one or more KPIs. The impact of the prominent time slot (special weekend) is analysed based on a decision tree comprising one or more impact probabilities, in one embodiment of the present disclosure.
[036] Further, system 100 applies a third data driven balancing technique amongst the plurality of data driven balancing techniques on the one or more KPIs. The third data driven balancing technique includes analysis of an impact (also referred to as impact analysis) of a prominent time slot on the one or more Key Performance Indicators (KPIs). The impact analysis is performed by the system 100 based on type of prominent time slot, a number of locations, one or more KPI affecting projects, and so on. For instance, the prominent time slot refers to a holiday, in the present disclosure. Therefore, the system 100 analyzes the impact of the holiday on the one or more KPIs. This impact analysis is performed by the system 100 based on the type of holiday, a number of locations, KPI associated projects affected, and historical data for similar scenario(s).
[037] Once the plurality of data driven balancing techniques on the one or more KPIs and the set of updated data driven balancing based KPIs are obtained, the set of updated data driven balancing based KPIs are analysed with respect to the historical data to obtain one or more time slots and one or more candidate KPI patterns. The system 100 then applies the machine learning model on at least one time interval (appropriate period) being selected based on the one or more time slots and the one or more candidate KPI patterns to obtain a predicted KPI pattern. The above steps are better understood by way of following description:
[038] For weekend data driven balancing (e.g., the first data driven balancing technique), first all the days of month for which KPIs to be forecasted with relevant past month are compared. Further, all necessary parameters such as number of days in month, number of working days, weekend difference, and/or the like are identified. These number of days in month, number of working days, weekend difference are referred to as one of more changing parameters as described above, furthermore, for different combination, the weekends data driven balancing is carried out. For each strategy for a month, all required actions are recorded and a logic of a function for each and every action is written, which are generally repetitive. Further, by moving distribution numbers according to target month, weekends are moved or balanced by the system 100, and a forecasted number provides better output after weekend data driven balancing. Post weekend data driven balancing, special weekends and holidays are then balanced.
[039] For special weekend data driven balancing (e.g., second data driven balancing technique), first the impact of special weekends on the KPIs (e.g., Billing) is analyzed by creating a decision tree that encompasses all the possibilities leading to an impact. Also, an impact using various factors related to weekends on month ends and historical data for similar scenario is analyzed. For instance, examples of various factors include, but are not limited to, (i) exceptional month(s), (ii) does the target month have a special weekend or not?, (iii) pattern, (iv) does the quarterly source month have special weekend or not?, (v) does the yearly source month have special weekend or not?, (vi) No. of days after the last weekend in the current year, (vii) Number of days after the last weekend in the previous year, (viii) Number of days after the last weekend in the previous quarter, Is it an exceptional month?, (ix) Are special weekend rules to be applied?, and the like. Further, for each leaf node in the decision tree, a strategy in the form of rules to adjust billing distribution percentage is defined which is further implemented. In other words, for each leaf node in the decision tree one or more rules are configured to adjust a KPI distribution percentage. The rules are depicted in FIG. 3, by way of examples. Based on pattern of the month, rules are determined by the system 100 of the present disclosure. However, depending upon the requirement and application of the present disclosure, in case of additional patterns observed in various and subsequent scenarios, rules can be added or modified for the same leaf node. Such exemplary rules shall not be construed as limiting the scope of the present disclosure.
[040] After special weekend data driven balancing, holiday data driven balancing (e.g., the third data driven balancing technique) is done by analyzing an impact of holiday on the billing is analyzed based on holiday type, number of locations, projects affected and historical data for similar scenario. Further, a strategy in the form of rules to adjust billing distribution percentage is defined which is further implemented.
[041] After holiday data driven balancing, data driven rebalanced (or readjusted) numbers are obtained which are again analyzed against historical data for similar leaf node pattern in the decision tree. Based on the identified similar leaf node pattern, weekend, month-end, weekday combination are identified, and a billing pattern is observed.
[042] Depending on data analysis, appropriate period is selected and artificial intelligence or machine learning based techniques are used to predict a billing pattern (e.g., candidate KPI pattern(s)) with good accuracy. Further, numbers are readjusted based on the predicted billing pattern (e.g., predicted KPI pattern). In other words, the system 100 readjusts the one or more KPIs based on the predicted KPI pattern to obtain one or more re-adjusted KPIs. The impact of billing done during a month is analyzed based on the predicted billing pattern and further numbers are continuously readjusted based on Billing done during the month. The process is repeated for calculating daily forecast by applying the predicted distribution to the forecasted number determined during month forecast estimation.
[043] The above step of 206 and various data driven balancing techniques as applied on the KPIs by the system 100 is further better understood by way of following description. Below Table 3 depicts details of KPI for a specific month Jan 2023:
Table 3
1 2 3 4 5 … 23 … 28 29 30 31 7 days
Actual Jan 2023 0 0 0 0 2 … 0 … 5 3 21 23 69
Rev pred Jan 0 0 1 1 3 … 3 … 5 3 19 23 66
Q Pred jan2023 0 0 1 0 0 … 0 … 21 8 3 15 67
Y Pred jan2023 0 0 1 1 3 … 3 … 9 4 15 23 66
Actual 01-2022 0 0 1 1 3 … 0 … 15 9 4 23 66
Actual 01-2021 0 0 0 1 1 … 2 … 20 24 1 0 63
Actual 01-2020 1 1 1 0 0 … 2 … 6 12 19 22 63
Actual 01-2019 1 1 2 0 0 … 2 … 8 15 18 20 68
Actual 01-2018 0 1 1 1 1 … 3 … 1 16 22 23 69
Actual 01-2017 0 1 0 0 0 … 2 … 7 4 21 29 86

[044] In the above Table 3, the numbers in the third row in bold (e.g., refer ‘Rev pred Jan’) are revised numbers from the column Y pred Jan 2023. These numbers are bold as previously mentioned business rules are applied on these numbers only. 9,4 are underlined in the fifth (5th) row because they signify unusual billing on special weekends, and 15 and 23 are Italic property because the underlined numbers affect them.
[045] Billing percentage numbers from the existing weekend data driven balancing algorithm are used to predict the numbers for the target month. For instance, two kinds of predictions are obtained: (i) Yearly predictions – These are obtained by using the billing percentage numbers from same month previous year of the target month, and (ii) Quarterly predictions – These are obtained by using the billing percentage numbers from same month previous quarter of the target month.
[046] The type of prediction is selected based on the historical data analysis. It is observed that if any of the days (Mon/Tue) after the weekend is not a holiday, the billing on Saturday of the weekend does not fall below 5%. Similarly, billing on Sunday does not rise above 3%. By this data-driven observation, in the above example, Saturday and Sunday are revised as 5% and 3% respectively. For the days after the weekend, 15% and 23% are revised to 19% and 23%. Mon (30th) - 4% is added to 30th (Monday) from Saturday since it is not a holiday and billing is observed to be more on non-holidays. This additional billing on 30th is justified because 15% does not fall in the billing range of 30th in the previous years. Tue (31st) - Since 23% falls in the range of the last day’s billing in the previous years on 31st, it was left unchanged. As shown in the above Table 3, this data driven balancing technique for January 2023 had given reasonably good predictions. However, it is to be understood by a person having ordinary skill in the art or person skilled in the art that the data driven balancing technique applied for the above-mentioned example scenario depicted in Table 3 may or may not be same for other examples in other leaf nodes. This may vary depending on the application and requirement of the entity.
[047] Below Table 4 depicts details of KPI for another specific month August 2021, wherein the system 100 applied various data driven balancing techniques as described above along with other steps of the method of FIG. 2:
Table 4
1 2 3 4 5 6 … 25 26 27 28 29 30 31 7 days
Actual
08-2021 0 2 2 2 2 2 … 5 8 12 5 4 14 16 64
Rev pred
Aug … 5 12 5 3 18 15 58
Y pred
Aug 2021 0 1 1 1 3 … 5 7 11 6 6 17 12 63
Q pred
Aug 2021 0 0 1 1 1 3 … 5 7 11 7 4 18 13 64
Actual
08-2020 0 0 1 1 1 3 … 5 7 11 17 6 6 12 63
Actual
08-2019 2 2 0 0 1 1 … 1 5 5 16 18 21 0 66
Actual
08-2018 5 1 1 0 0 0 … 1 0 8 12 16 15 17 69
Actual
08-2017 1 2 0 0 0 0 … 2 0 1 17 16 17 21 73
Actual
08-2016 1 0 0 0 0 0 … 9 10 2 1 22 25 22 91
Actual
08-2015 1 0 1 1 0 1 … 4 11 15 25 10 3 20 89

[048] As can be observed from Table 4, from 5th to last row, billing percentage numbers from past years are depicted. The third row depicts the quarterly prediction numbers obtained from weekend data driven balancing technique. The fourth row depicts the yearly prediction numbers obtained from weekend data driven balancing technique. The second row is the revised predictions obtained from applying the data driven balancing techniques, upon the predictions obtained from the weekend data driven balancing technique. Cells from 1-24 are left empty because revision is performed on the last 7 days unless there are major holidays. The first row depicts actuals of Aug 2021.
[049] As described above, the Billing percentage numbers from the weekend data driven balancing technique are used to predict the numbers for the target month. Two kinds of predictions are obtained (e.g., refer step 204): (i) Yearly predictions – These are obtained by using the billing percentage numbers from same month previous year of the target month, and (ii) Quarterly predictions – These are obtained by using the billing percentage numbers from same month previous quarter of the target month. The type of prediction is selected based on the historical data analysis (e.g., refer step 206 For this particular example, the system 100 selected yearly predictions, because there was a major holiday (Onam) on Aug 30th, 2020 (e.g., one type of holiday impact is observed here). The system 100 rejected quarterly predictions because of Onam. While applying the data driven balancing techniques, the system 100 observed that if any of the days (Mon/Tue) after the weekend is not a holiday, the billing on Saturday of the weekend does not fall below 5%. Similarly, billing on Sunday does not rise above 3% (e.g., refer step 206 and decision tree depicted in the form of Table 5). Referring to steps of FIG. 2 and the decision tree of Table 5, by this data-driven observation, in the above example, Saturday and Sunday are revised as 5% and 3% respectively. For the days after the weekend, 18% and 13% are revised to 18% and 15%. Tue (31st) - 2% is added to 31st (Tuesday) from Saturday since it is not a holiday and billing is observed to be more on non-holidays. Mon (30th) - Since 18% falls in the range of the last day’s billing in the previous years on 30th, it was left unchanged. For the days before the weekend, 1% from 28th was added to Friday (27th). This is in accordance with the billing range of the previous years. As shown in the above Table 4, the data driven balancing technique(s) applied for August 2021 by the system 100 and the method of the present disclosure has given reasonably good predictions. However, it is to be understood by a person having ordinary skill in the art or person skilled in the art that the data driven balancing technique applied for the above-mentioned example scenario depicted in Table 3 may or may not be same for other examples in other leaf nodes. This may vary depending on the application and requirement of the entity.
[050] The decision tree as mentioned above is better understood by way following description and examples and Table 5.
1. Consider Leaf node 1.
a. Condition#1
Target Month has a special weekend. Distribution pattern is Quarterly. Source month has special weekend. # of source month final weekdays equals to # of current month final weekdays
No Need to apply special weekend rules.
b. Condition#2
Target Month has a special weekend. Distribution pattern is Quarterly. Source month has special weekend. # of source month final weekdays does not equal to # of current month final weekdays.
Need to apply special weekend rules.
i. Sub-Condition A: no. of final weekdays=2. Monday is a holiday, but Tuesday is not.
Increase Saturday’s billing by the difference between old and new predictions on Monday and ensure that it is greater than ‘x%’ (e.g., 5%).
Similarly, ensure that Sunday's billing is less than or equal to ‘p%’ (e.g., 3%). This is applicable to both January 2023 and October 2023 and has been tested for both. The results are encouraging.
ii. Sub-Condition B: no. of final weekdays=2. Monday is not a holiday, but Tuesday is a holiday.
Decrease Saturday's billing by the difference between old and new predictions on Tuesday and ensure that it is greater than or equal to ‘x%’ (e.g., 5%). Similarly ensure that Sunday's billing is less than or equal to ‘p%’ (e.g., 3%).
iii. Sub-condition C: no. of final weekdays=2, both Monday and Tuesday are not holidays.
Increase Saturday's billing by the difference between old and new predictions on Monday and Decrease Sunday's billing by the difference between old and new predictions on Tuesday.
iv. Sub-condition D: no. of final weekdays=2. Both Monday and Tuesday are holidays.
No logic decided as no case observed so far.
The leaf node 1 with sub conditions is depicted in FIG. 3.
[051] More specifically, FIG. 3 depicts a portion of a decision tree with a condition leaf node illustrating various conditions and associated sub conditions as implemented by the system 100 of FIG. 1, in accordance with an embodiment of the present disclosure.
2. Consider Leaf node 2
a. Condition#1
Target Month has a special weekend. Distribution pattern is Quarterly. Source month does not have special weekend. Y source month has special weekend.
# of source month final weekdays Equals to # of current month final weekdays
No Need to apply special weekend rules.
b. Condition#2
Target Month has a special weekend. Distribution pattern is Quarterly. Source month does not have special weekend. Y source month has special weekend.
[052] FIGS. 4A through 4C, with reference to FIGS. 1 through 3, depict the decision tree as implemented by the system 100 of FIG. 1, in accordance with an embodiment of the present disclosure.
[053] While existing systems and methods provide common solutions such as use of preprocessing techniques, Neural Network and forecasting, these solutions have the following drawbacks such as forecasted output is prone to error due to change in distribution pattern, it does not provide prominent accuracy at day level. Embodiments of the present disclosure provide systems and methods that forecast KPIs for dataset of interest that pertain to one or more entities (e.g., organization, enterprise, and the like). More specifically, the system and method estimate monthly forecasting, wherein data from various dataset are read along with KPIs. The read data is preprocessed to ensure that it can be fitted into forecasting system/method of the present disclosure, wherein various strategies are applied to derive forecasted output. Embodiments of the present disclosure provide systems and methods that implement different data factors, such as special weekends, holidays, KPI projections, PESTLE (Political, Economic, Social, Technological, Legal and Environmental) factors, etc. to arrive at business rules/technique(s) that improves accuracy, derive data with similar patterns from existing data, and then use machine learning algorithms to arrive at a better prediction. The system and method analyze holidays due to festivals, government and public holidays and other types of holidays, in each country, that can potentially impact billing of respective customers from those countries. Different factors like political, economic, social, technological, legal and environment data can be collected with respect to billing geographic currency and then mapped with other data factors, which results in building business rules which can be used in prediction of billing. The system and method of the present disclosure observed/determined a strong correlation between billing projection and billing prediction. When incorporated by the system of the present disclosure, the prediction accuracy improved. In addition, the influence of KPI projection on KPI prediction has also been analyzed. In the present disclosure, the system 100 and the method implemented machine learning models such as SARIMA (Seasonal Auto-Regressive Integrated Moving Average with eXogenous), Facebook® prophet, long short-term memory (LSTM) for predicting KPIs for entities through various steps of FIG. 2. Depending on the application and requirements of the present disclosure any other machine learning model or artificial intelligence (AI) model may also be implemented.
[054] For addressing the above-mentioned technical problem, only machine learning solution is not enough. Additional data points are needed for analysis, that may positively or negatively impact billing. When the system 100 and the method of the present disclosure analyzed the transactional data, the format was not suitable to draw meaningful conclusion. Hence, the data was curated by the system 100 and format was changed wherein the detailed analysis presented challenges in deciding the right data driven balancing for redistributing billing percentage. This was finalized based on historical data analysis. Such analysis as implemented in the present disclosure can be applied to any industry like Energy, Transportation, Finance, Manufacturing, Media, and Entertainment, and so on. The key here is to create distribution and apply data driven balancing techniques for balancing datapoints as applicable (e.g., weekend(s), special weekends, holidays, and so on). Use cases like Billing, Energy Consumption, Production, Sales of Train tickets, etc. can be applied where weekend value is less. In addition to this use cases like sales of movie tickets, Sales of hotel rooms etc. can be applied when weekend value is prominent. This method can be applied on non-forecasting use case like copying and mapping a timetable from previous period to target period.
[055] The written description describes the subject matter herein to enable any person skilled in the art to make and use the embodiments. The scope of the subject matter embodiments is defined by the claims and may include other modifications that occur to those skilled in the art. Such other modifications are intended to be within the scope of the claims if they have similar elements that do not differ from the literal language of the claims or if they include equivalent elements with insubstantial differences from the literal language of the claims.
[056] It is to be understood that the scope of the protection is extended to such a program and in addition to a computer-readable means having a message therein; such computer-readable storage means contain program-code means for implementation of one or more steps of the method, when the program runs on a server or mobile device or any suitable programmable device. The hardware device can be any kind of device which can be programmed including e.g., any kind of computer like a server or a personal computer, or the like, or any combination thereof. The device may also include means which could be e.g., hardware means like e.g., an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, e.g., an ASIC and an FPGA, or at least one microprocessor and at least one memory with software processing components located therein. Thus, the means can include both hardware means and software means. The method embodiments described herein could be implemented in hardware and software. The device may also include software means. Alternatively, the embodiments may be implemented on different hardware devices, e.g., using a plurality of CPUs.
[057] The embodiments herein can comprise hardware and software elements. The embodiments that are implemented in software include but are not limited to, firmware, resident software, microcode, etc. The functions performed by various components described herein may be implemented in other components or combinations of other components. For the purposes of this description, a computer-usable or computer readable medium can be any apparatus that can comprise, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
[058] The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope of the disclosed embodiments. Also, the words “comprising,” “having,” “containing,” and “including,” and other similar forms are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items, or meant to be limited to only the listed item or items. It must also be noted that as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise.
[059] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer-readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.
[060] It is intended that the disclosure and examples be considered as exemplary only, with a true scope of disclosed embodiments being indicated by the following claims.
, Claims:

1. A processor implemented method, comprising:
obtaining, via one or more hardware processors, an input comprising one or more Key Performance Indicators (KPIs) specific to a first time period (202), wherein the one or more KPIs are associated with a dataset of interest pertaining to an entity;
predicting, via the one or more hardware processors, a distribution of the one or more KPIs in a plurality of time intervals comprised in the first time period based on a period specific forecast to obtain a set of predicted distribution of the one or more KPIs (204);
applying, via the one or more hardware processors, a plurality of data driven balancing techniques on the set of prediction distribution of the one or more KPIs to obtain a set of updated data driven balancing based KPIs (206),
wherein the step of applying a first data driven balancing technique amongst the plurality of data driven balancing techniques comprises:
performing a comparison of (i) each time slot in a current time interval amongst the plurality of time intervals for which the one or more Key Performance Indicators (KPIs) are to be forecasted and (ii) a corresponding time slot of a previous time interval; and
identifying one or more changing parameters for the current time interval based on the comparison,
wherein the step of applying a second data driven balancing technique amongst the plurality of data driven balancing techniques comprises:
analyzing an impact of a prominent time period on the one or more Key Performance Indicators (KPIs), and
wherein the step of applying a third data driven balancing technique amongst the plurality of data driven balancing techniques comprises:
analyzing an impact of a prominent time slot on the one or more KPIs based on a type of the prominent time slot, a number of locations, and one or more KPI affecting projects;
analyzing, via the one or more hardware processors, the set of updated data driven balancing based KPIs with respect to a historical data to obtain one or more time slots and one or more candidate KPI patterns (208); and
applying, via the one or more hardware processors, a machine learning model on at least one time interval being selected based on the one or more time slots and the one or more candidate KPI patterns to obtain a predicted KPI pattern (210).

2. The processor implemented method as claimed in claim 1, comprising re-adjusting the one or more KPIs based on the predicted KPI pattern to obtain one or more re-adjusted KPIs.

3. The processor implemented method as claimed in claim 1, wherein the impact of the prominent time slot is analysed based on a decision tree comprising one or more impact probabilities.

4. The processor implemented method as claimed in claim 3, wherein for each leaf node in the decision tree one or more rules are configured to adjust a KPI distribution percentage.

5. A system (100), comprising:
a memory (102) storing instructions;
one or more communication interfaces (106); and
one or more hardware processors (104) coupled to the memory (102) via the one or more communication interfaces (106), wherein the one or more hardware processors (104) are configured by the instructions to:
obtain an input comprising one or more Key Performance Indicators (KPIs) specific to a first time period, wherein the one or more KPIs are associated with a dataset of interest pertaining to an entity;
predict a distribution of the one or more KPIs in a plurality of time intervals comprised in the first time period based on a period specific forecast to obtain a set of predicted distribution of the one or more KPIs;
apply a plurality of data driven balancing techniques on the set of prediction distribution of the one or more KPIs to obtain a set of updated data driven balancing based KPIs,
wherein the step of applying a first data driven balancing technique amongst the plurality of data driven balancing techniques comprises:
performing a comparison of (i) each time slot in a current time interval amongst the plurality of time intervals for which the one or more Key Performance Indicators (KPIs) are to be forecasted and (ii) a corresponding time slot of a previous time interval; and
identifying one or more changing parameters for the current time interval based on the comparison,
wherein the step of applying a second data driven balancing technique amongst the plurality of data driven balancing techniques comprises:
analyzing an impact of a prominent time period on the one or more Key Performance Indicators (KPIs), and
wherein the step of applying a third data driven balancing technique amongst the plurality of data driven balancing techniques comprises:
analyzing an impact of a prominent time slot on the one or more KPIs based on a type of the prominent time slot, a number of locations, and one or more KPI affecting projects;
analyze the set of updated data driven balancing based KPIs with respect to a historical data to obtain one or more time slots and one or more candidate KPI patterns; and
apply a machine learning model on at least one time interval being selected based on the one or more time slots and the one or more candidate KPI patterns to obtain a predicted KPI pattern.

6. The system as claimed in claim 5, wherein the one or more hardware processors are further configured by the instructions to re-adjust the one or more KPIs based on the predicted KPI pattern to obtain one or more re-adjusted KPIs.

7. The system as claimed in claim 5, wherein the impact of the prominent time slot is analysed based on a decision tree comprising one or more impact probabilities.

8. The system as claimed in claim 7, wherein for each leaf node in the decision tree one or more rules are configured to adjust a KPI distribution percentage.