Add the score primitive which outputs scores of the predictions given the metrics
This is a primitive which takes in a predictions dataframe and also a score dataset with target columns. This primitive outputs a dataframe which stores the computed scores.
Before finalizing this primitive, there are some assumptions I make in writing this primitive:
-
First, this primitive currently takes in one dataframe and one dataset as input. Should we make it so that it takes in a list of prediction dataframe and also a list of datasets.
-
Second, I assume that only the main table contains target columns. Also, in computing the score, we need to extract the target columns from the input score dataset, currently I search the target columns in score dataset using the column name in input prediction dataframe. So the issues are as follows:
a. So this issue is about how we extract target columns from the score dataset and pair with the input predictions dataframe?
b. Although the input prediction dataframe may contain metadata, I currently uses the column name of the input dataframe to extract target column in score dataset. Is this process okay?