Predictive analytics factory

Abstract

Apparatuses, systems, methods, and computer program products are disclosed for a predictive analytics factory. A function generator module is configured to determine a plurality of learned functions based on training data without prior knowledge regarding suitability of the generated learned functions for the training data. A function evaluator module is configured to perform an evaluation of the plurality of learned functions using test data and to maintain evaluation metadata for the plurality of learned functions. A predictive compiler module is configured to form a predictive ensemble comprising a subset of multiple learned functions from the plurality of learned functions.

What is claimed is: 
     
         1 . An apparatus for a predictive analytics factory, the apparatus comprising:
 a function generator module configured to generate a plurality of learned functions based on training data without prior knowledge regarding suitability of the generated learned functions for the training data;   a function evaluator module configured to perform an evaluation of the plurality of learned functions using test data and to maintain evaluation metadata for the plurality of learned functions; and   a predictive compiler module configured to form a predictive ensemble based on the evaluation metadata, the predictive ensemble comprising a subset of multiple learned functions from the plurality of learned functions.   
     
     
         2 . The apparatus of  claim 1 , further comprising a feature selector module configured to, in response to the function generator module generating the plurality of learned functions, determine a subset of features from the training data for use in the predictive ensemble based on the evaluation metadata, the predictive compiler module configured to form the predictive ensemble using the selected subset of features. 
     
     
         3 . The apparatus of  claim 2 , wherein one or more of the features of the training data are selected by a user as required and the feature selector module is configured to select one or more optional features to include in the subset of features with the required one or more features. 
     
     
         4 . The apparatus of  claim 1 , wherein the predictive compiler module is configured to combine learned functions from the plurality of learned functions to form combined learned functions, the predictive ensemble comprising at least one combined learned function. 
     
     
         5 . The apparatus of  claim 1 , wherein the predictive compiler module is configured to add one or more layers to at least a portion of the plurality of learned functions to form one or more extended learned functions, at least one of the one or more layers comprising a probabilistic model, the predictive ensemble comprising at least one extended learned function. 
     
     
         6 . The apparatus of  claim 1 , wherein the predictive compiler module is configured to form the predictive ensemble by organizing the subset of multiple learned functions into the predictive ensemble, the predictive ensemble comprising the subset of multiple learned functions and a rule set synthesized from the evaluation metadata for the subset of learned functions to direct data through the multiple learned functions such that different learned functions of the ensemble process different subsets of the data based on the evaluation metadata. 
     
     
         7 . The apparatus of  claim 1 , further comprising an orchestration module configured to direct workload data through the predictive ensemble based on the evaluation metadata data to produce a classification for the workload data and a confidence metric for the classification. 
     
     
         8 . The apparatus of  claim 1 , further comprising an interface module configured to receive an analytics request from a client and to provide an analytics result to the client, the analytics request comprising workload data with similar features to the training data, the analytics result produced by the predictive ensemble. 
     
     
         9 . A system for a predictive analytics factory, the system comprising:
 a host computing device in communication with at least one client;   a predictive analytics module executing on the host computing device, the predictive analytics module determining a plurality of learned functions using training data received from the at least one client without prior knowledge regarding suitability of the determined learned functions for the training data, selecting a subset of the learned functions based on evaluation metadata generated for the plurality of learned functions, and forming a predictive ensemble comprising the selected subset of the learned functions from the plurality of learned functions.   
     
     
         10 . The system of  claim 9 , wherein the predictive analytics module comprises a predictive compiler and the plurality of learned function comprise computer readable code configured by the predictive compiler to accept an input comprising instances of one or more features of the training data and to provide a result. 
     
     
         11 . The system of  claim 10 , wherein the result comprises one or more of a classification, a confidence metric, an inferred function, a regression function, an answer, a prediction, a recognized pattern, a rule, a recommendation, a subset of the instances, and a subset of the one or more features. 
     
     
         12 . The system of  claim 9 , wherein the predictive analytics module generates and evaluates the plurality of learned functions using parallel computing on multiple processors of the host computing device. 
     
     
         13 . The system of  claim 9 , wherein the predictive analytics module determines the plurality of learned functions in response to a request from the at least one client, the request comprising a query, the ensemble formed to predict a result for the query. 
     
     
         14 . The system of  claim 13 , wherein the predictive analytics module returns the ensemble to the at least one client to satisfy the request. 
     
     
         15 . The system of  claim 13 , wherein the predictive analytics module returns the result for the query to the at least one client to satisfy the request, the predictive analytics module maintaining the ensemble in a library of a plurality of generated ensembles from which the predictive analytics module satisfies subsequent requests from the at least one client. 
     
     
         16 . The system of  claim 13 , wherein the predictive analytics module receives the request from the at least one client using one or more of an application programming interface, a shared library, a hardware command interface, and a data network. 
     
     
         17 . The system of  claim 9 , wherein the predictive ensemble comprises a rule set synthesized from the evaluation metadata to direct data through the subset of the learned functions. 
     
     
         18 . A predictive analytics ensemble comprising:
 multiple learned functions synthesized from a larger plurality of learned functions, the larger plurality of learned functions generated from training data without prior knowledge of a suitability of the larger plurality of learned functions for the training data;   a metadata rule set synthesized from the evaluation metadata for the plurality of learned functions for directing data through different learned functions of the multiple learned functions to produce a result; and   an orchestration module configured to direct the data through the different learned functions of the multiple learned functions based on the synthesized metadata rule set to produce the result.   
     
     
         19 . The predictive analytics ensemble of  claim 18 , further comprising a predictive correlation module configured to correlate one or more features of the multiple learned functions with a confidence metric associated with the result. 
     
     
         20 . The predictive analytics ensemble of  claim 19 , wherein the predictive correlation module is configured to provide a listing of the one or more features correlated with the result to a client.