Correlation Matrices (corr_matrix)

Correlation matrices can be computed using the corr_matrix function. The corr_matrix function takes two parameters:

  1. A string, enclosed in single quotes, containing a comma-separated list of numeric fields for which to calculate the matrix.

  2. The sample size to compute the correlation matrix from.

Sample syntax

select corr_matrix('petal_length_d, petal_width_d, sepal_length_d, sepal_width_d', 150) as corr,
       matrix_x,
       matrix_y
from iris

Result set

The result set for the corr_matrix function contains one row for each two field combination listed in the first parameter. The corr_matrix function returns the correlation for the two field combination. There are two additional fields, matrix_x and matrix_y that contain the field combination for the row.

Sample result set in Apache Zeppelin

Sample result

Visualization

The example below shows the corr_matrix result visualized in Apache Zeppelin with a heat map.

Sample visualization