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