df.col =pd.to_numeric(df.col, errors ='coerce').fillna(0).astype('int')
training_data['usagequantity'] = ( pd.to_numeric(training_data['usagequantity'], errors='coerce') .fillna(0) )