Spark Numeric Scorer

This node computes certain statistics between the a numeric column's values (r _i ) and predicted (p _i ) values. It computes R² =1-SS _res /SS _tot =1-Σ(p _i -r _i )²/Σ(r _i -1/n*Σr _i )² (can be negative!), mean absolute error (1/n*Σ|p _i -r _i |), mean squared error (1/n*Σ(p _i -r _i )²), root mean squared error (sqrt(1/n*Σ(p _i -r _i )²)), and mean signed difference (1/n*Σ(p _i -r _i )). The computed values can be inspected in the node's view and/or further processed using the output table.

Input ports

1. Type: Spark Data
   Table
   Arbitrary input Spark DataFrame/RDD with at least two numeric columns to compare.

Output ports

1. Type: Table
   Statistics
   The computed statistical measures:

   • R² - coefficient of determination , 1-SS_res/SS_tot
   • Mean squared error - 1/n*Σ((p_i-r_i)²)
   • Mean absolute error - 1/n*Σ|p_i-r_i|
   • Root mean squared error - Sqrt(1/n*Σ((p_i-r_i)²))
   • Mean signed difference - 1/n*Σ(p_i - r_i)