All Products
Search

This Product

    PolarDB:Pearson Correlation Coefficient

    Document Center

    PolarDB:Pearson Correlation Coefficient

    Last Updated:Mar 22, 2024

    This topic describes Pearson Correlation Coefficient that measures the linear correlation of two features. The larger the absolute value, the stronger the correlation.

    Scenarios

    Variables to which Pearson Correlation Coefficient can be applied must meet the following requirements:

    • The standard deviation of neither variable is 0.

    • The variables are in a linear relationship and are continuous.

    • The variables are in a bivariate normal distribution, or in a unimodal distribution that resembles a normal distribution.

    Pearson Correlation Coefficient is commonly used to determine the linear relationship of two features in a machine learning model. If two features are highly correlated, they may be interchangeable. In this case, you can discard one of them to ensure effectiveness of the model.

    Syntax

    CREATE FEATURE feature_name WITH ( feature_class = '', x_cols = '', parameters=()) AS (SELECT select_expr [, select_expr] ... FROM table_reference)

    Parameter description:

    Parameter

    Description

    Example

    feature_name

    The name of the feature.

    pearson_001

    feature_class

    The type of the feature. Set the value to pearson.

    pearson

    x_cols

    Custom parameters for creating the feature. Each value must be a floating point or an integer. Separate multiple variables with commas (,).

    dx1,dx2

    parameters

    Custom parameters for creating the feature. The following parameters are supported:

    • null_strategy: specifies how to replace NULL values. The following parameters are supported:

      • mean: replaces NULL values with the average value.

      • median: replaces NULL values with the median value.

    • categorical_feature: the categorical features. Separate multiple features with commas (,). Use this parameter to exclude non-integer and non-floating-point columns in x_cols.

    categorical_feature='dx3'

    select_expr

    The name of the column used to create the feature.

    dx4

    table_reference

    The name of the table containing the column used to create the feature.

    airlines_test_1000

    Example

    /*polar4ai*/CREATE FEATURE pearson_001 WITH ( feature_class = 'pearson',x_cols='Airline,Flight,AirportFrom,AirportTo,DayOfWeek,Time,Length',parameters=(null_strategy='mean',categorical_feature='Airline,Flight,AirportFrom,AirportTo,DayOfWeek')) AS (SELECT * FROM airlines_test_1000);