none
Query on "Filter based Features selection" function RRS feed

  • Question

  • Hello,

    I am new to MS Azure studio. I want to understand, how to use "Filter based Features selection" function to find the weightage/contribution of all columns in predicting a dependent variable value, for classification problems. Alternately, if there are other ways to get the weightage/contribution of all columns, you can share that too. 

    Also if there are sample experiments already available, which can study, then please share the sample experiments too.

    thanks.

    Sunday, January 12, 2020 6:00 PM

All replies

  • Hello KDR_Mumbai,

    Filter based Feature selection module helps us to identify the columns in your input dataset that have the greatest predictive power. 

    The Filter Based Feature Selection module provides multiple feature selection algorithms to choose from, including correlation methods such as Pearsons's or Kendall's correlation, mutual information scores, and chi-squared values. Azure Machine Learning also supports feature value counts as an indicator of information value.

    When you use the Filter Based Feature Selection module, you provide a dataset, identify the column that contains the label or dependent variable, and then specify a single method to use in measuring feature importance. The module outputs a dataset that contains the best feature columns, as ranked by predictive power. It also outputs the names of the features and their scores from the selected metric.

    Based on the available metrics that are chosen the output dataset varies. Please refer to the documentation on how the module can be configured based on the dataset you re using. 

    Sample experiments are available in the gallery if you need to run them before configuring them in your experiment.

    -----------------------------------------------------------------------------------------------------------
    If you found this post helpful, please give it a "Helpful" vote. 
    Please remember to mark the replies as answers if they help. 

    Tuesday, January 14, 2020 5:36 AM
    Moderator