69f972b898870734f36a8fd181fbc052037a9926,mindsdb/libs/phases/data_transformer/data_transformer.py,DataTransformer,run,#DataTransformer#Any#,91

Before Change

                        valid_rows = df[df[column] == val]

                        while max_val_occurances > len(valid_rows) + len(df):
                            df = df.append(valid_rows)
                        df = df.append(valid_rows[0:int(max_val_occurances - (len(valid_rows) + len(df)))])

After Change

                    else:
                        dfs = [input_data.train_df, input_data.test_df, input_data.validation_df]

                    for i in range(len(dfs)):
                        valid_rows = dfs[i][dfs[i][column] == val]

                        while max_val_occurances > len(valid_rows) + len(dfs[i]):
                            dfs[i] = dfs[i].append(valid_rows)
                        dfs[i] = dfs[i].append(valid_rows[0:int(max_val_occurances - (len(valid_rows) + len(dfs[i])))])

                print("\n\n-----------------\n\n")
                for df in [input_data.train_df, input_data.test_df, input_data.validation_df]:
                    print(len(df[df["default.payment.next.month"] == "0"]))
                    print(len(df[df["default.payment.next.month"] == "1"]))

Instances