PySpark: How to fillna values in dataframe for specific columns?

df.fillna(0, subset=['a', 'b'])

There is a parameter named subset to choose the columns unless your spark version is lower than 1.3.1