Prudential Life Insurance Assessment

This data set has few columns with extremely high number of null values. I don’t think there is any use in imputing such columns. Should I remove those columns altogether?

I think removing the column depends on it’s importance and it potential effect on the final prediction.