Rank Correlation

Calculates rank-based correlation coefficients between selected pairs of columns to measure the strength and direction of monotonic relationships.

This node supports several correlation methods based on ranked data:

• Spearman’s rank correlation : Measures the strength of a monotonic relationship between two variables. Values range from -1 (strong negative correlation) to +1 (strong positive correlation), with associated p-values and degrees of freedom computed.
• Goodman and Kruskal’s gamma : Assesses association based on concordant and discordant pairs; best suited for square contingency tables.
• Kendall’s tau : Includes Tau A and Tau B variants. Tau A ignores tied ranks, while Tau B adjusts for them and is more appropriate for rectangular tables.

The node uses fractional ranking for tied values. If column values lack a natural order, string representations are used for sorting. Columns of any data type can be analyzed, but results depend on the default ordering.

Rows with missing values are excluded from calculations. To apply different handling, address missing values beforehand.

Input ports

1. Type: Table
   Numeric input data
   Numeric input data to evaluate

Output ports

1. Type: Table
   Correlation measure
   Correlation variables, p-values and degrees of freedom.
2. Type: Table
   Correlation matrix
   Correlation variables in a matrix representation.
3. Type: Correlation
   Correlation model
   A model containing the correlation measures. This model is appropriate to be read by the Correlation Filter node.
4. Type: Table
   Rank table
   A table containing the fractional ranks of the columns. Where the rank corresponds to the values position in a sorted table.