Recommender system for tuning parameters to generate data analytics model and method thereof

A system for recommending tuning of parameters to generate a data analytics model, the system comprising:

• a recommender device comprising a processor and a memory communicatively coupled to the processor, wherein the memory stores processor-executable instructions, which, on execution, causes the processor to:
• identify via a first plurality of recommendation layers, at a data pre-processing stage for each of a plurality of pre-processing operations, a pre-processing subset from an associated set of predefined pre-processing methods for a predefined objective,
• wherein each pre-processing subset comprises a list of ranked predefined pre-processing methods,
• and wherein the pre-processing stage transforms input data into a set of pre-processed data;
• identify at a feature selection stage for each of a plurality of feature selection operations, a feature subset, comprising a list of ranked predefined feature selection methods, from an associated set of predefined feature selection methods for the predefined objective,
• wherein the identifying at the feature selection stage comprises:
• processing at the feature selection stage, through a second plurality of recommendation layers, the set of pre-processed data, a plurality of sets of predefined feature selection methods, and a list of problem types associated with the predefined objective, to generate a transformed data, wherein each set in the plurality of sets of predefined feature selection methods corresponds to a feature selection operation from a plurality of feature selection operations;
• assigning, for each of the plurality of feature selection operations, a rank to each predefined feature selection method in the associated feature subset based on the result of processing through each of the second plurality of recommendation layers;
• and
• generating for each of the plurality of feature selection operations, the feature subset from the associated set of predefined feature selection methods;
• identify at a model training stage for each of a plurality of model training operations, a training subset from an associated set of predefined model training methods for the predefined objective, wherein each training subset comprises a list of ranked predefined model training methods;
• generate a plurality of data analytics tuples, wherein each of the plurality of data analytics tuples comprises a predefined pre-processing method selected from the associated pre-processing subset, a predefined feature selection method selected from the associated feature subset, and a predefined model training method selected from the associated training subset;
• and
• select a data analytics tuple from the plurality of data analytics tuples, wherein an output result of the data analytics tuple comprises highest ranked results for the predefined objective, and wherein the data analytics tuple corresponds to the data analytics model.

A computer program product for recommending tuning of parameters to generate a data analytics model, the computer program product being embodied in a non-transitory computer readable storage medium of a recommender device and comprising computer instructions for:

• Identifying via a first plurality of recommendation layers, at a data pre-processing stage for each of a plurality of pre-processing operations, a pre-processing subset from an associated set of predefined pre-processing methods for a predefined objective,
• wherein each pre-processing subset comprises a list of ranked predefined pre-processing methods,
• and wherein the pre-processing stage transforms input data into a set of pre-processed data;
• identifying at a feature selection stage for each of a plurality of feature selection operations, a feature subset, comprising a list of ranked predefined feature selection methods, from an associated set of predefined feature selection methods for the predefined objective,
• wherein the identifying at the feature selection stage comprises:
• processing at the feature selection stage, through a second plurality of recommendation layers, the set of pre-processed data, a plurality of sets of predefined feature selection methods, and a list of problem types associated with the predefined objective, to generate a transformed data, wherein each set in the plurality of sets of predefined feature selection methods corresponds to a feature selection operation from a plurality of feature selection operations;
• assigning, for each of the plurality of feature selection operations, a rank to each predefined feature selection method in the associated feature subset based on the result of processing through each of the second plurality of recommendation layers;
• and
• generating for each of the plurality of feature selection operations, the feature subset from the associated set of predefined feature selection methods;
• identifying at a model training stage for each of a plurality of model training operations, a training subset from an associated set of predefined model training methods for the predefined objective, wherein each training subset comprises a list of ranked predefined model training methods;
• generating a plurality of data analytics tuples, wherein each of the plurality of data analytics tuples comprises a predefined pre-processing method selected from the associated pre-processing subset, a predefined feature selection method selected from the associated feature subset, and a predefined model training method selected from the associated training subset;
• and
• selecting a data analytics tuple from the plurality of data analytics tuples, wherein an output result of the data analytics tuple comprises highest ranked results for the predefined objective, and wherein the data analytics tuple corresponds to the data analytics model.