| 41 | # so only call fit() once |
| 42 | # call transform() for any subsequent data |
| 43 | class DataTransformer: |
| 44 | def fit(self, df): |
| 45 | self.scalers = {} |
| 46 | for col in NUMERICAL_COLS: |
| 47 | scaler = StandardScaler() |
| 48 | scaler.fit(df[col].values.reshape(-1, 1)) |
| 49 | self.scalers[col] = scaler |
| 50 | |
| 51 | def transform(self, df): |
| 52 | N, _ = df.shape |
| 53 | D = len(NUMERICAL_COLS) + len(NO_TRANSFORM) |
| 54 | X = np.zeros((N, D)) |
| 55 | i = 0 |
| 56 | for col, scaler in iteritems(self.scalers): |
| 57 | X[:,i] = scaler.transform(df[col].values.reshape(-1, 1)).flatten() |
| 58 | i += 1 |
| 59 | for col in NO_TRANSFORM: |
| 60 | X[:,i] = df[col] |
| 61 | i += 1 |
| 62 | return X |
| 63 | |
| 64 | def fit_transform(self, df): |
| 65 | self.fit(df) |
| 66 | return self.transform(df) |
| 67 | |
| 68 | |
| 69 | def get_data(): |