System and method to build and score predictive model for numerical attributes

1. A system to build and score predictive model for numerical attributes comprising:

a processing subsystem; and

a memory coupled to the processing subsystem, wherein the memory comprises a set of program instructions, configured to be executed by the processing subsystem, wherein the processing subsystem is configured to:

select one or more numerical variables from a plurality of data sets that are acquired from one or more sources, based on a plurality of parameters, wherein the plurality of parameters comprises at least one of a use case, a statistical influence and a previous predictive sample;

extract a plurality of features from the plurality of data sets based on feature engineering and transformation processes applied on the one or more numerical variables;

perform one of a time series forecasting and a regression technique on the one or more numerical variables based on the plurality of features extracted, wherein performing one of the time series forecasting and the regression technique on the one or more numerical variables comprises:

performing the time series forecasting on the one or more numerical variables when the one or more numerical variables of the plurality of data sets are in a time series;

perform the regression technique on the one or more numerical variables when the one or more numerical variables of the plurality of data sets are not in said time series;

evaluate and select a prediction technique using the regression technique based on a plurality of elements, wherein the plurality of elements comprises at least one of a data quantity, a data volume, a computational resource, a data type, a use case, a plurality of features, a plurality of model performance and a historical model;

generate a prediction model based on the selected prediction technique using one of the time series forecasting and the regression technique, wherein the generated prediction model is split into a training prediction model for self-learning and a test prediction model, wherein the training prediction model is tested for accuracy upon self-learning of the training prediction model;

compare the training prediction model with an accuracy criteria;

calculate a score for the generated prediction model when the training prediction model matches with the accuracy criteria by automatically crawling through external sources;

predict the one or more numerical variables based on the calculated score for the generated prediction model; and

present a predicted result of the generated prediction model in one or more forms, wherein the one or more forms comprise a graph, a chart, a table or an insight wherein, the insight is a textual insight in a natural language.

2. The system of claim 1 , wherein the plurality of data sets is acquired from at least one of a web, a manual entry of data, a local data set, an internal storage, an external storage and an experimental data set.

3. A method for scoring and predicting numerical data comprising:

acquiring, by a processing subsystem, a plurality of data sets from one or more sources;

selecting, by the processing subsystem, one or more numerical variables from the plurality of data sets based on a plurality of parameters, wherein the plurality of parameters comprises at least one of a use case, a statistical influence and a previous predictive sample;

extracting, by the processing subsystem, a plurality of features from the plurality of data sets based on feature engineering and transformation processes applied on the one or more numerical variables;

performing, by the processing subsystem, one of a time series forecasting and a regression technique on the one or more numerical variables based on the plurality of features extracted, wherein performing the time series forecasting on the one or more numerical variables comprises:

performing, by the processing subsystem, the time series forecasting on the one or more numerical variables when the one or more numerical variables of the plurality of data sets are in a time series;

performing, by the processing subsystem, the regression technique on the one or more numerical variables when the one or more numerical variables of the plurality of data sets are not in said time series;

evaluating and selecting, by the processing subsystem, a prediction technique using the regression technique based on a plurality of elements, wherein the plurality of elements comprises at least one of a data quantity, a data volume, a computational resource, a data type, a use case, a plurality of features, a plurality of model performance and a historical model;

generating, by the processing subsystem, a prediction model based on the selected prediction technique using one of the time series forecasting and the regression technique, wherein the generated prediction model is split into a training prediction model for self-learning and a test prediction model, wherein the training prediction model is tested for accuracy upon self-learning of the training prediction model;

comparing, by the processing subsystem, the training prediction model with an accuracy criteria;

calculating, by the processing subsystem, a score for the generated prediction model when the training prediction model matches with the accuracy criteria by automatically crawling through external sources;

predicting, by the processing subsystem, the one or more numerical variables based on the calculated score for the generated prediction model; and

presenting, by a processing subsystem, a predicted result of the generated prediction model in one or more forms, wherein the one or more forms comprise a graph, a chart, a table or an insight wherein, the insight is a textual insight in a natural language.

4. The method of claim 3 , wherein the plurality of data sets is acquired from at least one of a web, a manual entry of data, a local data set, an internal storage, an external storage and an experimental data set.