Claims:1. A processor-implemented method for forecasting using combined estimators, comprising:
providing a multivariate data as an input data to a data analysis module (302);
formatting the input data (304);
identifying a set of patterns and priorities from the formatted input data (306);
predicting missing gaps in the formatted input data (308);
merging the predicted missing data with the formatted input data to generate a complete dataset using the identified set of patterns and priorities of the formatted input data (310);
performing a plurality of forecasting techniques on the complete dataset to generate a plurality of forecasting models, wherein each of the plurality of forecasting models independently forecasts to generate a plurality of forecasted simulation results (312);
comparing the plurality of forecasted simulation results with each other based on combined estimation techniques to generate a plurality of decisions (314);
optimizing the plurality of decisions based on the plurality of forecasted simulation results (316); and
displaying the optimized plurality of decisions for further (318).

2. The method of claim 1, wherein the step of performing forecasting includes short-term and long-term forecasting, wherein the duration of short-term and long-term forecasting is pre-defined.

3. The method of claim 1, wherein the multivariate input data is formatted using data fusion techniques.

4. The method of claim 1, wherein techniques used to identify set of patterns and priorities from the formatted input data include Box-Cox method, Markov chain Monte Carlo (MCMC) simulation method, Bayesian structural time series (BSTS) method, Bagging, Boosting, Bayesian moving average, best available technology (BATS) method and Autoregressive integrated moving average ( ARIMA ).

5. The method of claim 1, wherein the missing gaps in the formatted input data is predicted and merged to formatted input data based on identified patterns-priorities of the formatted input data of at least past two years using a pre-defined arbitrary percentage value.

6. The method of claim 1, wherein plurality of forecasting techniques are one or more of:
Autoregressive integrated moving average (ARIMA) model forecasting,
TBATS (T for trigonometric regressors to model multiple-seasonalities, B for Box-Cox transformations, A for ARIMA errors, T for trend, S for seasonality) model forecasting,
error forecasting, trend and seasonal (ETS) model forecasting, and
Bayesian structural time series (BSTS) model forecasting.

7. The method of claim 5, wherein each of the plurality of forecasting models independently forecasts and visually simulates the forecasted results for the user to view plurality of forecasted simulation results.

8. The method of claim 1, wherein the estimation techniques for step of comparison include one or more of Bagging Algorithm, Boosting Algorithm and Bayesian Model Averaging.

9. The method of claim 1, wherein optimizing plurality of decisions is based on optimizing weighted average of the forecasted results of multiple combined estimates taken, with the weights being the inverse of the variances of each of the forecasted results of combined estimates.

10. A system for forecasting using combined estimators, the system (100) comprising:
a memory (102) storing instructions and one or more modules (108);
one or more communication or input/output 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 execute the one or more modules (108) comprising:
a data analysis module (202) for receiving a multivariate data as an input data;
a formatting module(204) for formatting the input data;
an identification module (206) for:
identifying a set of patterns and priorities from the formatted input data,
predicting missing gaps in the formatted input data in the data analysis module (202), and
merging the predicted missing data with the formatted input data to generate a complete dataset using the identified set of patterns and priorities of the formatted input data in the data analysis module (202);
a forecasting module (208) for performing a plurality of forecasting techniques on the complete dataset to generate a plurality of forecasting models, wherein each of the plurality of forecasting models independently forecasts to generate a plurality of forecasted simulation results;
a combined estimation module (216) for comparing the plurality of forecasted simulation results with each other based on an estimation technique to generate a plurality of decisions;
an optimized decision engine (218) for optimizing the plurality of decisions based on the plurality of forecasted simulation results; and
a display module (220) for displaying the optimized plurality of decisions for further analysis.
, 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:

METHOD AND SYSTEM FOR FORECASTING USING COMBINED ESTIMATORS

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

The disclosure herein generally relates to forecasting methods and systems and, more particularly, to a system and a method for forecasting using combined estimators.

BACKGROUND

Forecasting is a process of predicting future based on past/historic and present data. Forecasting is one of the prime management activities in every organization that helps in identifying future demand patterns and facilitates development to address the forecasted demand. Hence, forecasting is an integral part of decision-making activities of every organization. Further, forecasting is ubiquitous throughout many domains including chemical, utility, automotive industries, transportation and so on.
Organizations require short-term and long-term forecasts, depending on application. Short-term forecasts are required for scheduling of personnel, production and transportation requirements while long-term forecasts are implemented in strategic planning for decision-making to improve activities in an organization by considering account of market opportunities, environmental factors and internal resources.
Due to growing complexity in data, accurate forecasting is uncertain. Further complex large data may have missing gap (scarce data) which is also a major challenge for accurate forecasting. Forecasting also requires development of expertise in identifying/ analyzing forecasting problems, applying a range of forecasting methods, selecting appropriate methods for each problem, and evaluating and refining forecasting methods over time. Hence analyzing data accurately and choosing the right forecasting technique or choosing a combination of forecasting techniques would enable efficient forecasting even for most complex data, with missing gaps.

SUMMARY

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. For example, in one embodiment, a system and a method for forecasting using combined estimators. The proposed forecasting using combined estimators is a single unified combined estimating platform for forecasting, wherein a variety of forecasting techniques have been incorporated into a single unified forecasting platform to provide for accurate forecasting. Further the proposed single unified forecasting platform can work efficiently even when data available is scare, wherein input data includes missing gaps/data.
In another aspect, a method for forecasting using combined estimators is disclosed. The method further includes providing a multivariate data as an input data to a data analysis module. Further the method includes formatting the input data. The method further includes identifying a set of patterns and priorities from the formatted input data. Further the method includes predicting missing gaps in the formatted input data. The method further includes merging the predicted missing data with the formatted input data to generate a complete dataset using the identified set of patterns and priorities of the formatted input data. Further the method includes performing a plurality of forecasting techniques on the complete dataset to generate a plurality of forecasting models, wherein each of the plurality of forecasting models independently forecasts to generate a plurality of forecasted simulation results. The method further includes comparing the plurality of forecasted simulation results with each other based on combined estimation techniques to generate a plurality of decisions. Further the method includes optimizing the plurality of decisions based on the plurality of forecasted simulation results and displaying the optimized plurality of decisions for further analysis.
In yet another aspect, a system for forecasting using combined estimators, the system comprising a memory storing instructions and one or more modules, 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 execute the one or more modules further comprising a data analysis module for receiving a multivariate data as an input data. Further the system comprises of a formatting module for formatting the input data. The system further comprises an identification module for identifying a set of patterns and priorities from the formatted input data, predicting missing gaps in the formatted input data in the data analysis module, and merging the predicted missing data with the formatted input data to generate a complete dataset using the identified set of patterns and priorities of the formatted input data in the data analysis module. Further the system comprises of a forecasting modules for performing a plurality of forecasting techniques on the complete dataset to generate a plurality of forecasting models, wherein each of the plurality of forecasting models independently forecasts to generate a plurality of forecasted simulation results. The system further comprises a combined estimation module for comparing the plurality of forecasted simulation results with each other based on an estimation technique to generate a plurality of decisions. Further the system comprises of an optimized decision engine for optimizing the plurality of decisions based on the plurality of forecasted simulation results and display module for displaying the optimized plurality of decisions for further analysis.
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

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:
FIG. 1 illustrates an exemplary block diagram of a system for forecasting using combined estimators in accordance with some embodiments of the present disclosure.
FIG. 2 is a functional block diagram of various modules stored in module(s) of a memory of the system of FIG. 1 in accordance with some embodiments of the present disclosure.
FIG. 3 is an exemplary flow diagram illustrating a method for forecasting using combined estimators using the system of FIG. 1 in accordance with some embodiments of the present disclosure.
FIG 4A, 4B and 4C is an exemplary graph illustrating forecasted results for the user to view plurality of forecasted simulation results, in accordance with some embodiments of the present disclosure.
FIG 5A and 5B is an exemplary graph illustrating the results of plurality of decisions to result in a single estimate and error verification table, in accordance with some embodiments of the present disclosure.

DETAILED DESCRIPTION OF EMBODIMENTS

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 spirit and scope of the disclosed embodiments. It is intended that the following detailed description be considered as exemplary only, with the true scope and spirit being indicated by the following claims.
Referring now to the drawings, and more particularly to FIGS.1 through FIGS.5 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.
FIG. 1 illustrates an exemplary block diagram of a system 100 for forecasting using combined estimators according to an embodiment of the present disclosure. In an embodiment, the system 100 includes one or more processors 102, communication interface device(s) or input/output (I/O) interface(s) 104, and one or more data storage devices or memory 106 operatively coupled to the one or more processors 102. The memory 106 comprises one or more modules 108 and the database 110. The one or more processors 102 that are 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 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, workstations, mainframe computers, servers, a network cloud and the like.
The I/O interface device(s) 104 can include a variety of software and hardware interfaces, for example, a web interface, a graphical subject 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.
The memory 106 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.
FIG. 2, with reference to FIG. 1, is a block diagram of various modules 108 stored in the memory 106 of the system 100 of FIG. 1 in accordance with an embodiment of the present disclosure. In an embodiment of the present disclosure, the module 108 is represented as system 200 in FIG.2, comprises a data analysis module 202 that further comprises of a formatting module 204 and an identification module 206. Further the system 200 in FIG.2, comprises a forecasting modules 208, a combined estimation module 216, an optimized decision engine 218 and a display module 220, wherein the said modules are implemented as at least one of a logically self-contained part of a software program, a self-contained hardware component, and/or, a self-contained hardware component with a logically self-contained part of a software program embedded into each of the hardware component that when executed perform the above method described herein.
According to an embodiment of the disclosure, the system 200 comprises a data analysis module 202 that is configured for receiving a multivariate data as an input data. The multivariate data involves multiple variables and in an embodiment, a multivariate input data taken from a Sugarcane forecasting process includes sugarcane data (origin, country, length, state, sugar content, pin code, etc.), weather data, soil data, and season data.
According to an embodiment of the disclosure, the system 200 further comprises a formatting module 204 within data analysis module 202. The formatting module 204 is configured to format the multivariate input data. The multivariate input data is formatted using formatting techniques that include data fusion techniques.
According to an embodiment of the disclosure, the system 200 further comprises identification module 206 within repositioning module 204. The identification module 206 is configured to perform a variety of functions that include identifying a set of patterns and priorities from the formatted input data, wherein techniques used to identify set of patterns and priorities from the formatted input data include Box-Cox method, Markov chain Monte Carlo (MCMC) simulation method, Bayesian structural time series (BSTS) method, Bagging, Boosting, Bayesian moving average, best available technology (BATS) method and Autoregressive integrated moving average (ARIMA). Further the identification module 206 is also configured for predicting missing gaps in the formatted input data and further merging the predicted missing data with the formatted input data to generate a complete dataset. The missing gaps are predicted based on the identified set of patterns and priorities of the formatted input data.
In an embodiment, missing gaps predicted for a daily volume sales data considered from January to December of a year is explained. Given the data for third year, the missing patterns (second year and first year) are predicted as shown below;
The data for third year is represented as Y_3 (t) (1)
The data for second year is represented as Y_2 (t) (2)
The data for first year is represented as Y_1 (t) (3)
Further to predicted missing gaps based on a pre-defined arbitrary percentage value to ensure the best-suited model in terms of error as shown below;
Y_2 (t)=a%(Y_2 (t)) (4)
Y_1 (t)=ß%(Y_3 (t))|ß%(Y_2 (t)) (5)
Where a and ß are arbitrary values.
In an embodiment, the arbitrary values are chosen as 5 and 10, however for the purpose of brevity, the value of a and ß can be taken in the range {1, 10}, to the fluctuating demand in the market. Further, the forecasting may be done for a horizon of one month (thirty one days) on a daily basis, however the horizon values may also be either increased or decreased based on fluctuating demand in the market.
The missing gaps are predicted before performing forecasting techniques to ensure that forecasting modules 208 analyze, visualize and forecast in an accurate manner.
According to an embodiment of the disclosure, the system 200 further comprises forecasting modules 208, wherein the forecasting modules 208 further comprises of a plurality of individual forecasting modules that includes forecasting module-1 210, forecasting module-2 212 and forecasting module-N 212. The forecasting modules 208 are configured to perform a plurality of forecasting techniques on the complete dataset to generate a plurality of forecasting models, wherein each of the plurality of forecasting models independently forecasts to generate a plurality of forecasted simulation results. Further the model to be used forecasting is decided based on the type of forecasting required such as for long term forecasting where past data availability is scarce , qualitative forecasting techniques such as delphi method, historical life-cycle analogy are performed. Further for short term forecasting, quantitative forecasting models such as weighted N period moving average, simple exponential smoothing, Poisson process model based forecasting are used to forecast future data as a function of past data, as it is reasonable to assume that some of patterns in the data are expected to continue into the future. Further few other forecasting techniques that may be performed on the complete dataset include Autoregressive integrated moving average (ARIMA) model forecasting, TBATS (T for trigonometric regressors to model multiple-seasonalities, B for Box-Cox transformations, A for ARIMA errors, T for trend, S for seasonality) model forecasting, error forecasting, trend and seasonal (ETS) model forecasting, and Bayesian structural time series (BSTS) model forecasting. Further each of the plurality of forecasting models independently forecasts and visually simulates the forecasted results for the user to view plurality of forecasted simulation results, as shown in FIG 4A, 4B and 4C.
According to an embodiment of the disclosure, the system 200 further comprises a combined estimation module 216 .The a combined estimation module 216 is configured for performing estimation techniques for step of comparison include one or more of Bagging Algorithm, Boosting Algorithm and Bayesian Model Averaging.
According to an embodiment of the disclosure, the system 200 further comprises an optimized decision engine 218. The optimized decision engine 218 is configured for optimizing plurality of decisions to result in a single estimate based on optimizing weighted average of the forecasted results of multiple combined estimates taken, with the weights being the inverse of the variances of each of the forecasted results of combined estimates which is explained using the expressions below,
Let X_1, X_2,.. X_i represent plurality of decisions, with means µ_1, µ_2 . . µ_i, and with varianceV_1, V_2, … V_i.
To get maximum precision of best estimate, the variance must be minimized as shown below;
?Y=t_1 X?_1+ t_1 X_2+?= ?¦?t_i X_i ? (6)
As each of the estimates is unbiased,
?¦t_i =1 (7)
As the X_iare independent,
var(Y)=?¦?t_i^2 v_i ? (8)
To get maximum precision, Var(Y) must be minimized. Further using lagrange multipliers for this constrained optimization, and differentiating with respect to t_1
t_i=k/v_i (9)
?¦?t_i=1? (10)
k=(??¦v_i^(-1) )?^(-1)) (11)
Y_opt=k?¦X_i/v_i (12)
E[Y_opt ]= k?¦u_i (13)
var(Y_opt )= t_i^2 v_i (14)
=k^2 ?¦1/v_i (15)
=k (16)
var(Y_opt )=(??¦v_i^(-1) )?^(-1) (17)

In an embodiment, a graph illustrating the results of plurality of decisions to result in a single estimate and error verification table is shown in FIG 5A and FIG. 5B respectively, wherein it can been seen that estimated best estimator yield a much better result instead of working separately as individual plurality of forecasting models.
According to an embodiment of the disclosure, the system 200 further comprises a display module 220 .The a display module 220 is configured for displaying the optimized plurality of decisions for further analysis, based on genetic algorithm decision support tool, that can analyze information from the input multivariate data first and then the results from all the models including combined estimators.
FIG. 3, with reference to FIGS. 1-2, is an exemplary flow diagram illustrating a method for forecasting using combined estimators using the system 100 of FIG. 1, according to an embodiment of the present disclosure. In an embodiment, the system 100 comprises one or more data storage devices or the memory 106 operatively coupled to the one or more hardware processors 102 and is configured to store instructions for execution of steps of the method by the one or more processors 102. The steps of the method of the present disclosure will now be explained with reference to the components of the system 100 and the modules 202-220 as depicted in FIGS. 1-2, and the flow diagram.
At step 302, multivariate data is provided as input data to a data analysis module 202.
In the next step at 304, the input data is formatted in a formatting module 204, wherein the multivariate input data is formatted using data fusion techniques.
In the next step at 306, a set of patterns and priorities are identified from the formatted input data in an identification module 206. Further techniques used to identify set of patterns and priorities from the formatted input data include Box-Cox method, Markov chain Monte Carlo (MCMC) simulation method, Bayesian structural time series (BSTS) method, Bagging, Boosting, Bayesian moving average, best available technology (BATS) method and Autoregressive integrated moving average ( ARIMA ).
In the next step at 308, predicting missing gaps in the formatted input data in an identification module (206) based on identified patterns-priorities of the formatted input data of at least past two years using a pre-defined arbitrary percentage value.
In the next step at 310, the predicted missing data is merged with the formatted input data to generate a complete dataset using the identified set of patterns and priorities of the formatted input data in an identification module 206.
In the next step at 312, a plurality of forecasting techniques are performed on the complete dataset to generate a plurality of forecasting models, wherein each of the plurality of forecasting models independently forecasts to generate a plurality of forecasted simulation results in a forecasting modules 208. The plurality of forecasting techniques performed on the complete dataset include Autoregressive integrated moving average (ARIMA) model forecasting, TBATS (T for trigonometric regressors to model multiple-seasonalities, B for Box-Cox transformations, A for ARIMA errors, T for trend, S for seasonality) model forecasting, error forecasting, trend and seasonal (ETS) model forecasting, and Bayesian structural time series (BSTS) model forecasting. Further each of the plurality of forecasting models independently forecasts and visually simulates the forecasted results for the user to view plurality of forecasted simulation results.
In the next step at 314, the plurality of forecasted simulation results are comparing with each other based on combined estimation techniques to generate a plurality of decisions in a combined estimation module 216. The estimation techniques for step of comparison include one or more of Bagging Algorithm, Boosting Algorithm and Bayesian Model Averaging.
In the next step at 316, the plurality of decisions are optimizing based on the plurality of forecasted simulation results in a combined estimation module 216 based on optimizing weighted average of the forecasted results of multiple combined estimates taken, with the weights being the inverse of the variances of each of the forecasted results of combined estimates.
In the next step at 318, the optimized plurality of decisions is displayed in a display module 220 for further analysis.
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.
Hence a system and method for forecasting using combined estimators is provided. The proposed forecasting using combined estimators is a single unified combined estimating platform for forecasting, wherein a variety of forecasting techniques have been incorporated into a single unified forecasting platform to provide for accurate forecasting. Further the proposed single unified forecasting platform can work efficiently even when data available is scare, wherein input data includes missing gaps/data.
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 modules 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.
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 modules described herein may be implemented in other modules or combinations of other modules. 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.
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 and spirit 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.
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.
It is intended that the disclosure and examples be considered as exemplary only, with a true scope and spirit of disclosed embodiments being indicated by the following claims.