How to use udf functions in pyspark

Question

I am analysing the following piece of code:

from pyspark.sql.functions import udf,col, desc    
def MAE(y_true, y_pred):
    return abs(y_true - y_pred)

udf_