Skip to content

stemflow.model.special_hurdle

LightGBM_Hurdle

Bases: BaseEstimator

A simple Hurdle model class

Source code in stemflow/model/special_hurdle.py 
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
class LightGBM_Hurdle(BaseEstimator):
    """A simple Hurdle model class"""

    def __init__(
        self,
        classifier_params: dict = {
            "objective": "binary",
            "metric": ["auc", "binary_logloss"],
            "num_threads": 1,
            "verbosity": -1,
        },
        regressor_params: dict = {"objective": "regression", "metric": ["rmse"], "num_threads": 1, "verbosity": -1},
    ):
        """Make a LightGBM Hurdle class object

        Args:
            classifier_params:
                params for LightGBM for classification tasks
            regressor_params:
                params for LightGBM for regression tasks


        Example:
            ```py
            >>> model = AdaSTEMRegressor(
                    base_model=LightGBM_Hurdle(),
                    save_gridding_plot = True,
                    ensemble_fold=10,
                    min_ensemble_required=7,
                    grid_len_lon_upper_threshold=25,
                    grid_len_lon_lower_threshold=5,
                    grid_len_lat_upper_threshold=25,
                    grid_len_lat_lower_threshold=5,
                    points_lower_threshold=50,
                    Spatio1='longitude',
                    Spatio2 = 'latitude',
                    Temporal1 = 'DOY',
                    use_temporal_to_train=True,
                    n_jobs=1
                )

            >>> # fit adastem
            >>> model.fit(X_train.reset_index(drop=True), y_train, verbosity=0)

            >>> ...

            ```

        """

        self.classifier_params = classifier_params
        self.regressor_params = regressor_params
        self.classifier = None
        self.regressor = None

    def fit(self, X_train: Union[pd.core.frame.DataFrame, np.ndarray], y_train: Sequence, sample_weight=None):
        """Fitting model

        Args:
            X_train:
                Training variables
            y_train:
                Training target

        """
        binary_ = np.unique(np.where(y_train > 0, 1, 0))

        if len(binary_) == 1:
            warnings.warn("Warning: only one class presented. Replace with dummy classifier & regressor.")
            self.classifier = dummy_model1(binary_[0])
            self.regressor = dummy_model1(binary_[0])
            return

        # cls
        cls_dat = lgb.Dataset(np.array(X_train), label=np.where(np.array(y_train).flatten() > 0, 1, 0))

        if sample_weight is not None:
            cls_dat.set_weight(sample_weight)
        else:
            pass

        cls_ = lgb.train(self.classifier_params, cls_dat)
        self.classifier = cls_

        # reg
        reg_dat = lgb.Dataset(
            np.array(X_train)[np.array(y_train).flatten() > 0, :],
            label=np.array(y_train).flatten()[np.array(y_train).flatten() > 0],
        )
        reg_ = lgb.train(self.regressor_params, reg_dat)
        self.regressor = reg_

        # try:
        #     self.feature_importances_ = (np.array(self.classifier.feature_importances_) + np.array(self.regressor.feature_importances_))/2
        # except Exception as e:
        #     pass

    def predict(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
        """Predicting

        Args:
            X_test: Test variables

        Returns:
            A prediction array with shape (-1,1)
        """
        cls_res = self.classifier.predict(X_test)
        cls_res = np.where(cls_res > 0.5, 1, cls_res)
        cls_res = np.where(cls_res < 0.5, 0, cls_res)
        reg_res = self.regressor.predict(X_test)
        # reg_res = np.where(reg_res>=0, reg_res, 0) # we constrain the reg value to be positive
        res = np.where(cls_res > 0, reg_res, cls_res)
        return res.reshape(-1, 1)

    def predict_proba(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
        """Predicting probability

        This method output a numpy array with shape (n_sample, 2)
        However, user should notice that this is only for structuring the sklearn predict_proba-like method
        Only the res[:,1] is meaningful, aka the last dimension in the two dimensions. The first dimension is always zero.

        Args:
            X_test:
                Testing variables

        Returns:
            Prediction results with shape (n_samples, 2)
        """
        a = np.zeros(len(X_test)).reshape(-1, 1)
        b = self.predict(X_test).reshape(-1, 1)
        res = np.concatenate([a, b], axis=1)
        return res

__init__(classifier_params={'objective': 'binary', 'metric': ['auc', 'binary_logloss'], 'num_threads': 1, 'verbosity': -1}, regressor_params={'objective': 'regression', 'metric': ['rmse'], 'num_threads': 1, 'verbosity': -1})

Make a LightGBM Hurdle class object

Parameters:

  • classifier_params (dict, default: {'objective': 'binary', 'metric': ['auc', 'binary_logloss'], 'num_threads': 1, 'verbosity': -1} ) –

    params for LightGBM for classification tasks

  • regressor_params (dict, default: {'objective': 'regression', 'metric': ['rmse'], 'num_threads': 1, 'verbosity': -1} ) –

    params for LightGBM for regression tasks

Example 
>>> model = AdaSTEMRegressor(
        base_model=LightGBM_Hurdle(),
        save_gridding_plot = True,
        ensemble_fold=10,
        min_ensemble_required=7,
        grid_len_lon_upper_threshold=25,
        grid_len_lon_lower_threshold=5,
        grid_len_lat_upper_threshold=25,
        grid_len_lat_lower_threshold=5,
        points_lower_threshold=50,
        Spatio1='longitude',
        Spatio2 = 'latitude',
        Temporal1 = 'DOY',
        use_temporal_to_train=True,
        n_jobs=1
    )

>>> # fit adastem
>>> model.fit(X_train.reset_index(drop=True), y_train, verbosity=0)

>>> ...
Source code in stemflow/model/special_hurdle.py 
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
def __init__(
    self,
    classifier_params: dict = {
        "objective": "binary",
        "metric": ["auc", "binary_logloss"],
        "num_threads": 1,
        "verbosity": -1,
    },
    regressor_params: dict = {"objective": "regression", "metric": ["rmse"], "num_threads": 1, "verbosity": -1},
):
    """Make a LightGBM Hurdle class object

    Args:
        classifier_params:
            params for LightGBM for classification tasks
        regressor_params:
            params for LightGBM for regression tasks


    Example:
        ```py
        >>> model = AdaSTEMRegressor(
                base_model=LightGBM_Hurdle(),
                save_gridding_plot = True,
                ensemble_fold=10,
                min_ensemble_required=7,
                grid_len_lon_upper_threshold=25,
                grid_len_lon_lower_threshold=5,
                grid_len_lat_upper_threshold=25,
                grid_len_lat_lower_threshold=5,
                points_lower_threshold=50,
                Spatio1='longitude',
                Spatio2 = 'latitude',
                Temporal1 = 'DOY',
                use_temporal_to_train=True,
                n_jobs=1
            )

        >>> # fit adastem
        >>> model.fit(X_train.reset_index(drop=True), y_train, verbosity=0)

        >>> ...

        ```

    """

    self.classifier_params = classifier_params
    self.regressor_params = regressor_params
    self.classifier = None
    self.regressor = None

fit(X_train, y_train, sample_weight=None)

Fitting model

Parameters:

  • X_train (Union[DataFrame, ndarray]) –

    Training variables

  • y_train (Sequence) –

    Training target

Source code in stemflow/model/special_hurdle.py 
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
def fit(self, X_train: Union[pd.core.frame.DataFrame, np.ndarray], y_train: Sequence, sample_weight=None):
    """Fitting model

    Args:
        X_train:
            Training variables
        y_train:
            Training target

    """
    binary_ = np.unique(np.where(y_train > 0, 1, 0))

    if len(binary_) == 1:
        warnings.warn("Warning: only one class presented. Replace with dummy classifier & regressor.")
        self.classifier = dummy_model1(binary_[0])
        self.regressor = dummy_model1(binary_[0])
        return

    # cls
    cls_dat = lgb.Dataset(np.array(X_train), label=np.where(np.array(y_train).flatten() > 0, 1, 0))

    if sample_weight is not None:
        cls_dat.set_weight(sample_weight)
    else:
        pass

    cls_ = lgb.train(self.classifier_params, cls_dat)
    self.classifier = cls_

    # reg
    reg_dat = lgb.Dataset(
        np.array(X_train)[np.array(y_train).flatten() > 0, :],
        label=np.array(y_train).flatten()[np.array(y_train).flatten() > 0],
    )
    reg_ = lgb.train(self.regressor_params, reg_dat)
    self.regressor = reg_

predict(X_test)

Predicting

Parameters:

  • X_test (Union[DataFrame, ndarray]) –

    Test variables

Returns:

  • ndarray –

    A prediction array with shape (-1,1)

Source code in stemflow/model/special_hurdle.py 
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
def predict(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
    """Predicting

    Args:
        X_test: Test variables

    Returns:
        A prediction array with shape (-1,1)
    """
    cls_res = self.classifier.predict(X_test)
    cls_res = np.where(cls_res > 0.5, 1, cls_res)
    cls_res = np.where(cls_res < 0.5, 0, cls_res)
    reg_res = self.regressor.predict(X_test)
    # reg_res = np.where(reg_res>=0, reg_res, 0) # we constrain the reg value to be positive
    res = np.where(cls_res > 0, reg_res, cls_res)
    return res.reshape(-1, 1)

predict_proba(X_test)

Predicting probability

This method output a numpy array with shape (n_sample, 2) However, user should notice that this is only for structuring the sklearn predict_proba-like method Only the res[:,1] is meaningful, aka the last dimension in the two dimensions. The first dimension is always zero.

Parameters:

  • X_test (Union[DataFrame, ndarray]) –

    Testing variables

Returns:

  • ndarray –

    Prediction results with shape (n_samples, 2)

Source code in stemflow/model/special_hurdle.py 
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
def predict_proba(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
    """Predicting probability

    This method output a numpy array with shape (n_sample, 2)
    However, user should notice that this is only for structuring the sklearn predict_proba-like method
    Only the res[:,1] is meaningful, aka the last dimension in the two dimensions. The first dimension is always zero.

    Args:
        X_test:
            Testing variables

    Returns:
        Prediction results with shape (n_samples, 2)
    """
    a = np.zeros(len(X_test)).reshape(-1, 1)
    b = self.predict(X_test).reshape(-1, 1)
    res = np.concatenate([a, b], axis=1)
    return res

XGBoost_Hurdle

Bases: BaseEstimator

A simple Hurdle model class

Source code in stemflow/model/special_hurdle.py 
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
class XGBoost_Hurdle(BaseEstimator):
    """A simple Hurdle model class"""

    def __init__(
        self,
        classifier_params: dict = {
            "objective": "binary:logistic",
            "metric": ["auc", "logloss"],
            "tree_method": "hist",
            "nthread": 1,
            "verbosity": 0,
        },
        regressor_params: dict = {
            "objective": "reg:squarederror",
            "metric": ["rmse"],
            "tree_method": "hist",
            "nthread": 1,
            "verbosity": 0,
        },
    ):
        """Make a XGBoost Hurdle class object

        Args:
            classifier_params:
                params for XGBoost for classification tasks
            regressor_params:
                params for XGBoost for regression tasks

        Example:
            ```py
            >>> model = AdaSTEMRegressor(
                    base_model=XGBoost_Hurdle(),
                    save_gridding_plot = True,
                    ensemble_fold=10,
                    min_ensemble_required=7,
                    grid_len_lon_upper_threshold=25,
                    grid_len_lon_lower_threshold=5,
                    grid_len_lat_upper_threshold=25,
                    grid_len_lat_lower_threshold=5,
                    points_lower_threshold=50,
                    Spatio1='longitude',
                    Spatio2 = 'latitude',
                    Temporal1 = 'DOY',
                    use_temporal_to_train=True,
                    n_jobs=1
                )

            >>> # fit adastem
            >>> model.fit(X_train.reset_index(drop=True), y_train, verbosity=0)

            >>> ...

            ```

        """
        self.classifier_params = classifier_params
        self.regressor_params = regressor_params
        self.classifier = None
        self.regressor = None

    def fit(self, X_train: Union[pd.core.frame.DataFrame, np.ndarray], y_train: Sequence, sample_weight=None):
        """Fitting model

        Args:
            X_train:
                Training variables
            y_train:
                Training target

        """
        binary_ = np.unique(np.where(y_train > 0, 1, 0))

        if len(binary_) == 1:
            warnings.warn("Warning: only one class presented. Replace with dummy classifier & regressor.")
            self.classifier = dummy_model1(binary_[0])
            self.regressor = dummy_model1(binary_[0])
            return

        # cls
        if sample_weight is not None:
            cls_dat = xgb.DMatrix(
                np.array(X_train), label=np.where(np.array(y_train).flatten() > 0, 1, 0), weight=sample_weight
            )
        else:
            cls_dat = xgb.DMatrix(np.array(X_train), label=np.where(np.array(y_train).flatten() > 0, 1, 0))

        cls_ = xgb.train(self.classifier_params, cls_dat)
        self.classifier = cls_

        # reg
        reg_dat = xgb.DMatrix(
            np.array(X_train)[np.array(y_train).flatten() > 0, :],
            label=np.array(y_train).flatten()[np.array(y_train).flatten() > 0],
        )
        reg_ = xgb.train(self.regressor_params, reg_dat)
        self.regressor = reg_

        # try:
        #     self.feature_importances_ = (np.array(self.classifier.feature_importances_) + np.array(self.regressor.feature_importances_))/2
        # except Exception as e:
        #     pass

    def predict(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
        """Predicting

        Args:
            X_test: Test variables

        Returns:
            A prediction array with shape (-1,1)
        """
        if isinstance(self.classifier, dummy_model1):
            res = np.array([self.classifier.the_value] * X_test.shape[0])
        else:
            X_test_dmat = xgb.DMatrix(np.array(X_test))
            cls_res = self.classifier.predict(X_test_dmat)
            cls_res = np.where(cls_res > 0.5, 1, cls_res)
            cls_res = np.where(cls_res < 0.5, 0, cls_res)
            reg_res = self.regressor.predict(X_test_dmat)
            res = np.where(cls_res > 0, reg_res, cls_res)

        return res.reshape(-1, 1)

    def predict_proba(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
        """Predicting probability

        This method output a numpy array with shape (n_sample, 2)
        However, user should notice that this is only for structuring the sklearn predict_proba-like method
        Only the res[:,1] is meaningful, aka the last dimension in the two dimensions. The first dimension is always zero.

        Args:
            X_test:
                Testing variables

        Returns:
            Prediction results with shape (n_samples, 2)
        """
        a = np.zeros(len(X_test)).reshape(-1, 1)
        b = self.predict(X_test).reshape(-1, 1)
        res = np.concatenate([a, b], axis=1)
        return res

__init__(classifier_params={'objective': 'binary:logistic', 'metric': ['auc', 'logloss'], 'tree_method': 'hist', 'nthread': 1, 'verbosity': 0}, regressor_params={'objective': 'reg:squarederror', 'metric': ['rmse'], 'tree_method': 'hist', 'nthread': 1, 'verbosity': 0})

Make a XGBoost Hurdle class object

Parameters:

  • classifier_params (dict, default: {'objective': 'binary:logistic', 'metric': ['auc', 'logloss'], 'tree_method': 'hist', 'nthread': 1, 'verbosity': 0} ) –

    params for XGBoost for classification tasks

  • regressor_params (dict, default: {'objective': 'reg:squarederror', 'metric': ['rmse'], 'tree_method': 'hist', 'nthread': 1, 'verbosity': 0} ) –

    params for XGBoost for regression tasks

Example 
>>> model = AdaSTEMRegressor(
        base_model=XGBoost_Hurdle(),
        save_gridding_plot = True,
        ensemble_fold=10,
        min_ensemble_required=7,
        grid_len_lon_upper_threshold=25,
        grid_len_lon_lower_threshold=5,
        grid_len_lat_upper_threshold=25,
        grid_len_lat_lower_threshold=5,
        points_lower_threshold=50,
        Spatio1='longitude',
        Spatio2 = 'latitude',
        Temporal1 = 'DOY',
        use_temporal_to_train=True,
        n_jobs=1
    )

>>> # fit adastem
>>> model.fit(X_train.reset_index(drop=True), y_train, verbosity=0)

>>> ...
Source code in stemflow/model/special_hurdle.py 
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
def __init__(
    self,
    classifier_params: dict = {
        "objective": "binary:logistic",
        "metric": ["auc", "logloss"],
        "tree_method": "hist",
        "nthread": 1,
        "verbosity": 0,
    },
    regressor_params: dict = {
        "objective": "reg:squarederror",
        "metric": ["rmse"],
        "tree_method": "hist",
        "nthread": 1,
        "verbosity": 0,
    },
):
    """Make a XGBoost Hurdle class object

    Args:
        classifier_params:
            params for XGBoost for classification tasks
        regressor_params:
            params for XGBoost for regression tasks

    Example:
        ```py
        >>> model = AdaSTEMRegressor(
                base_model=XGBoost_Hurdle(),
                save_gridding_plot = True,
                ensemble_fold=10,
                min_ensemble_required=7,
                grid_len_lon_upper_threshold=25,
                grid_len_lon_lower_threshold=5,
                grid_len_lat_upper_threshold=25,
                grid_len_lat_lower_threshold=5,
                points_lower_threshold=50,
                Spatio1='longitude',
                Spatio2 = 'latitude',
                Temporal1 = 'DOY',
                use_temporal_to_train=True,
                n_jobs=1
            )

        >>> # fit adastem
        >>> model.fit(X_train.reset_index(drop=True), y_train, verbosity=0)

        >>> ...

        ```

    """
    self.classifier_params = classifier_params
    self.regressor_params = regressor_params
    self.classifier = None
    self.regressor = None

fit(X_train, y_train, sample_weight=None)

Fitting model

Parameters:

  • X_train (Union[DataFrame, ndarray]) –

    Training variables

  • y_train (Sequence) –

    Training target

Source code in stemflow/model/special_hurdle.py 
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
def fit(self, X_train: Union[pd.core.frame.DataFrame, np.ndarray], y_train: Sequence, sample_weight=None):
    """Fitting model

    Args:
        X_train:
            Training variables
        y_train:
            Training target

    """
    binary_ = np.unique(np.where(y_train > 0, 1, 0))

    if len(binary_) == 1:
        warnings.warn("Warning: only one class presented. Replace with dummy classifier & regressor.")
        self.classifier = dummy_model1(binary_[0])
        self.regressor = dummy_model1(binary_[0])
        return

    # cls
    if sample_weight is not None:
        cls_dat = xgb.DMatrix(
            np.array(X_train), label=np.where(np.array(y_train).flatten() > 0, 1, 0), weight=sample_weight
        )
    else:
        cls_dat = xgb.DMatrix(np.array(X_train), label=np.where(np.array(y_train).flatten() > 0, 1, 0))

    cls_ = xgb.train(self.classifier_params, cls_dat)
    self.classifier = cls_

    # reg
    reg_dat = xgb.DMatrix(
        np.array(X_train)[np.array(y_train).flatten() > 0, :],
        label=np.array(y_train).flatten()[np.array(y_train).flatten() > 0],
    )
    reg_ = xgb.train(self.regressor_params, reg_dat)
    self.regressor = reg_

predict(X_test)

Predicting

Parameters:

  • X_test (Union[DataFrame, ndarray]) –

    Test variables

Returns:

  • ndarray –

    A prediction array with shape (-1,1)

Source code in stemflow/model/special_hurdle.py 
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
def predict(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
    """Predicting

    Args:
        X_test: Test variables

    Returns:
        A prediction array with shape (-1,1)
    """
    if isinstance(self.classifier, dummy_model1):
        res = np.array([self.classifier.the_value] * X_test.shape[0])
    else:
        X_test_dmat = xgb.DMatrix(np.array(X_test))
        cls_res = self.classifier.predict(X_test_dmat)
        cls_res = np.where(cls_res > 0.5, 1, cls_res)
        cls_res = np.where(cls_res < 0.5, 0, cls_res)
        reg_res = self.regressor.predict(X_test_dmat)
        res = np.where(cls_res > 0, reg_res, cls_res)

    return res.reshape(-1, 1)

predict_proba(X_test)

Predicting probability

This method output a numpy array with shape (n_sample, 2) However, user should notice that this is only for structuring the sklearn predict_proba-like method Only the res[:,1] is meaningful, aka the last dimension in the two dimensions. The first dimension is always zero.

Parameters:

  • X_test (Union[DataFrame, ndarray]) –

    Testing variables

Returns:

  • ndarray –

    Prediction results with shape (n_samples, 2)

Source code in stemflow/model/special_hurdle.py 
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
def predict_proba(self, X_test: Union[pd.core.frame.DataFrame, np.ndarray]) -> np.ndarray:
    """Predicting probability

    This method output a numpy array with shape (n_sample, 2)
    However, user should notice that this is only for structuring the sklearn predict_proba-like method
    Only the res[:,1] is meaningful, aka the last dimension in the two dimensions. The first dimension is always zero.

    Args:
        X_test:
            Testing variables

    Returns:
        Prediction results with shape (n_samples, 2)
    """
    a = np.zeros(len(X_test)).reshape(-1, 1)
    b = self.predict(X_test).reshape(-1, 1)
    res = np.concatenate([a, b], axis=1)
    return res