I'am ML newbie but I decided to create a program in ML.NET for data classification. However, my dataset has a very large number of columns, so I'll use another example.

There is an iris database with five columns:

• sepalLength (number)
• sepalWidth (number)
• petalWidth (number)
• petalWidth (number)
• irisClass (predicted -> Iris-setosa/Iris-versicolor/Iris-virginica)

Is it possible to investigate to what extent a given data column influences the classification of a flower to a given type of iris? For example, the value of the sepalLength column influences the result by 60%, while the value of the petalWidth column influences the result by 5%.

CodePudding user response：

I think what you're looking for is either permutation feature importance or feature contribution calculation