()
| 67 | |
| 68 | |
| 69 | def get_data(): |
| 70 | df = pd.read_csv('housing.data', header=None, delim_whitespace=True) |
| 71 | df.columns = [ |
| 72 | 'crim', # numerical |
| 73 | 'zn', # numerical |
| 74 | 'nonretail', # numerical |
| 75 | 'river', # binary |
| 76 | 'nox', # numerical |
| 77 | 'rooms', # numerical |
| 78 | 'age', # numerical |
| 79 | 'dis', # numerical |
| 80 | 'rad', # numerical |
| 81 | 'tax', # numerical |
| 82 | 'ptratio', # numerical |
| 83 | 'b', # numerical |
| 84 | 'lstat', # numerical |
| 85 | 'medv', # numerical -- this is the target |
| 86 | ] |
| 87 | |
| 88 | # transform the data |
| 89 | transformer = DataTransformer() |
| 90 | |
| 91 | # shuffle the data |
| 92 | N = len(df) |
| 93 | train_idx = np.random.choice(N, size=int(0.7*N), replace=False) |
| 94 | test_idx = [i for i in range(N) if i not in train_idx] |
| 95 | df_train = df.loc[train_idx] |
| 96 | df_test = df.loc[test_idx] |
| 97 | |
| 98 | Xtrain = transformer.fit_transform(df_train) |
| 99 | Ytrain = np.log(df_train['medv'].values) |
| 100 | Xtest = transformer.transform(df_test) |
| 101 | Ytest = np.log(df_test['medv'].values) |
| 102 | return Xtrain, Ytrain, Xtest, Ytest |
| 103 | |
| 104 | |
| 105 | if __name__ == '__main__': |
no test coverage detected