Feature selection method and system for regression analysis / model construction
AssigneeAT&T Intellectual Property I, L.P.
InventorsVladimir Sevastyanov, Abhay Dabholkar, Rakhi Gupta, James H. Pratt, Nikhlesh Agrawal
Life of this patent
Estimating the term…
Abstract
Aspects of the subject disclosure may include, for example, dividing a feature range of a feature into a plurality of subsets that span the feature range, calculating an average target variable value for each subset of the plurality of subsets, resulting in a plurality of average target variable values, and estimating a measure of feature significance with respect to a target variable by determining a difference between a maximum average target variable value and a minimum average target variable value in the plurality of average target variable values. Other embodiments are disclosed.
Claims20
A method, comprising:
- selecting, by a processing system including a processor, a feature from a plurality of features in a data set for generating a machine learning model, wherein:
- the plurality of features are determined to potentially impact a target variable of the machine learning model, and
- the data set comprises feature values of the plurality of features associated with target variable values of the target variable;
- dividing, by the processing system, a feature range of the feature into a plurality of subsets that span the feature range, wherein the feature range comprises a value range of the feature values for the feature from the data set;
- calculating, by the processing system, an average target variable value from the target variable values associated with the feature values for each subset of the plurality of subsets, resulting in a plurality of average target variable values;
- estimating, by the processing system, a measure of feature significance with respect to the target variable by determining a difference between a maximum average target variable value and a minimum average target variable value in the plurality of average target variable values;
- and
- generating the machine learning model using at least the feature based on the measure of feature significance indicating an impact on the target variable.
A device, comprising:
- a processing system including a processor;
- and
- a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations, the operations comprising:
- determining a type of each feature of a plurality of features associated with a target variable in a data set for generating a machine learning model, wherein:
- the plurality of features are determined to potentially impact a target variable of the machine learning model, and
- the data set comprises feature values of the plurality of features associated with target variable values of the target variable;
- for each feature of the plurality of features, estimating a respective significance value for that feature, relative to the target variable, based on the type of that feature and the feature values and the target variable values of that feature, resulting in a plurality of respective significant values;
- performing filtering of the plurality of features based on the plurality of respective significant values;
- and
- generating the machine learning model using a set of features that were filtered from the plurality of features.
A non-transitory machine-readable medium, comprising executable instructions that, when executed by a processing system including a processor, facilitate performance of operations, the operations comprising:
- identifying a plurality of subsets for a feature in a data set, wherein the data set comprises:
- a plurality of features that are determined to potentially impact a target variable of the machine learning model, and
- feature values of the plurality of features associated with target variable values of the target variable;
- determining an average target variable value from the target variable values associated with the feature for each subset of the plurality of subsets, resulting in a plurality of average target variable values;
- calculating a difference between a maximum average target variable value and a minimum average target variable value in the plurality of average target variable values, wherein the difference corresponds to a measure of feature significance with respect to the target variable;
- and
- based on the measure of feature significance satisfying a threshold, utilizing the feature to construct a regression or predictive model.
Dependencies are inferred from the claim text.