Skip to content

Commit 5eb7d62

Browse files
Copilotthinkall
Copilot
and
thinkall
authored
Expand docs to include all flamlized estimators (#1472)
* Initial plan * Add documentation for all flamlized estimators (RandomForest, ExtraTrees, LGBMClassifier, XGBRegressor) Co-authored-by: thinkall <3197038+thinkall@users.noreply.github.com> * Fix markdown formatting per pre-commit Co-authored-by: thinkall <3197038+thinkall@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: thinkall <3197038+thinkall@users.noreply.github.com> Co-authored-by: Li Jiang <bnujli@gmail.com>
1 parent 22dcfcd commit 5eb7d62

1 file changed

Lines changed: 232 additions & 0 deletions

File tree

‎website/docs/Examples/Default-Flamlized.md‎

Lines changed: 232 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -67,6 +67,82 @@ X_test.shape: (5160, 8), y_test.shape: (5160,)
6767

6868
[Link to notebook](https://github.com/microsoft/FLAML/blob/main/notebook/zeroshot_lightgbm.ipynb) | [Open in colab](https://colab.research.google.com/github/microsoft/FLAML/blob/main/notebook/zeroshot_lightgbm.ipynb)
6969

70+
## Flamlized LGBMClassifier
71+
72+
### Prerequisites
73+
74+
This example requires the [autozero] option.
75+
76+
```bash
77+
pip install flaml[autozero] lightgbm openml
78+
```
79+
80+
### Zero-shot AutoML
81+
82+
```python
83+
from flaml.automl.data import load_openml_dataset
84+
from flaml.default import LGBMClassifier
85+
from flaml.automl.ml import sklearn_metric_loss_score
86+
87+
X_train, X_test, y_train, y_test = load_openml_dataset(dataset_id=1169, data_dir="./")
88+
lgbm = LGBMClassifier()
89+
lgbm.fit(X_train, y_train)
90+
y_pred = lgbm.predict(X_test)
91+
print(
92+
"flamlized lgbm accuracy",
93+
"=",
94+
1 - sklearn_metric_loss_score("accuracy", y_pred, y_test),
95+
)
96+
print(lgbm)
97+
```
98+
99+
#### Sample output
100+
101+
```
102+
load dataset from ./openml_ds1169.pkl
103+
Dataset name: airlines
104+
X_train.shape: (404537, 7), y_train.shape: (404537,);
105+
X_test.shape: (134846, 7), y_test.shape: (134846,)
106+
flamlized lgbm accuracy = 0.6745
107+
LGBMClassifier(colsample_bytree=0.85, learning_rate=0.05, max_bin=255,
108+
min_child_samples=20, n_estimators=500, num_leaves=31,
109+
reg_alpha=0.01, reg_lambda=0.1, verbose=-1)
110+
```
111+
112+
## Flamlized XGBRegressor
113+
114+
### Prerequisites
115+
116+
This example requires xgboost, sklearn, openml==0.10.2.
117+
118+
### Zero-shot AutoML
119+
120+
```python
121+
from flaml.automl.data import load_openml_dataset
122+
from flaml.default import XGBRegressor
123+
from flaml.automl.ml import sklearn_metric_loss_score
124+
125+
X_train, X_test, y_train, y_test = load_openml_dataset(dataset_id=537, data_dir="./")
126+
xgb = XGBRegressor()
127+
xgb.fit(X_train, y_train)
128+
y_pred = xgb.predict(X_test)
129+
print("flamlized xgb r2", "=", 1 - sklearn_metric_loss_score("r2", y_pred, y_test))
130+
print(xgb)
131+
```
132+
133+
#### Sample output
134+
135+
```
136+
load dataset from ./openml_ds537.pkl
137+
Dataset name: houses
138+
X_train.shape: (15480, 8), y_train.shape: (15480,);
139+
X_test.shape: (5160, 8), y_test.shape: (5160,)
140+
flamlized xgb r2 = 0.8542
141+
XGBRegressor(colsample_bylevel=1, colsample_bytree=0.85, learning_rate=0.05,
142+
max_depth=6, n_estimators=500, reg_alpha=0.01, reg_lambda=1.0,
143+
subsample=0.9)
144+
```
145+
70146
## Flamlized XGBClassifier
71147

72148
### Prerequisites
@@ -112,3 +188,159 @@ XGBClassifier(base_score=0.5, booster='gbtree',
112188
scale_pos_weight=1, subsample=1.0, tree_method='hist',
113189
use_label_encoder=False, validate_parameters=1, verbosity=0)
114190
```
191+
192+
## Flamlized RandomForestRegressor
193+
194+
### Prerequisites
195+
196+
This example requires the [autozero] option.
197+
198+
```bash
199+
pip install flaml[autozero] scikit-learn openml
200+
```
201+
202+
### Zero-shot AutoML
203+
204+
```python
205+
from flaml.automl.data import load_openml_dataset
206+
from flaml.default import RandomForestRegressor
207+
from flaml.automl.ml import sklearn_metric_loss_score
208+
209+
X_train, X_test, y_train, y_test = load_openml_dataset(dataset_id=537, data_dir="./")
210+
rf = RandomForestRegressor()
211+
rf.fit(X_train, y_train)
212+
y_pred = rf.predict(X_test)
213+
print("flamlized rf r2", "=", 1 - sklearn_metric_loss_score("r2", y_pred, y_test))
214+
print(rf)
215+
```
216+
217+
#### Sample output
218+
219+
```
220+
load dataset from ./openml_ds537.pkl
221+
Dataset name: houses
222+
X_train.shape: (15480, 8), y_train.shape: (15480,);
223+
X_test.shape: (5160, 8), y_test.shape: (5160,)
224+
flamlized rf r2 = 0.8521
225+
RandomForestRegressor(max_features=0.8, min_samples_leaf=2, min_samples_split=5,
226+
n_estimators=500)
227+
```
228+
229+
## Flamlized RandomForestClassifier
230+
231+
### Prerequisites
232+
233+
This example requires the [autozero] option.
234+
235+
```bash
236+
pip install flaml[autozero] scikit-learn openml
237+
```
238+
239+
### Zero-shot AutoML
240+
241+
```python
242+
from flaml.automl.data import load_openml_dataset
243+
from flaml.default import RandomForestClassifier
244+
from flaml.automl.ml import sklearn_metric_loss_score
245+
246+
X_train, X_test, y_train, y_test = load_openml_dataset(dataset_id=1169, data_dir="./")
247+
rf = RandomForestClassifier()
248+
rf.fit(X_train, y_train)
249+
y_pred = rf.predict(X_test)
250+
print(
251+
"flamlized rf accuracy",
252+
"=",
253+
1 - sklearn_metric_loss_score("accuracy", y_pred, y_test),
254+
)
255+
print(rf)
256+
```
257+
258+
#### Sample output
259+
260+
```
261+
load dataset from ./openml_ds1169.pkl
262+
Dataset name: airlines
263+
X_train.shape: (404537, 7), y_train.shape: (404537,);
264+
X_test.shape: (134846, 7), y_test.shape: (134846,)
265+
flamlized rf accuracy = 0.6701
266+
RandomForestClassifier(max_features=0.7, min_samples_leaf=3, min_samples_split=5,
267+
n_estimators=500)
268+
```
269+
270+
## Flamlized ExtraTreesRegressor
271+
272+
### Prerequisites
273+
274+
This example requires the [autozero] option.
275+
276+
```bash
277+
pip install flaml[autozero] scikit-learn openml
278+
```
279+
280+
### Zero-shot AutoML
281+
282+
```python
283+
from flaml.automl.data import load_openml_dataset
284+
from flaml.default import ExtraTreesRegressor
285+
from flaml.automl.ml import sklearn_metric_loss_score
286+
287+
X_train, X_test, y_train, y_test = load_openml_dataset(dataset_id=537, data_dir="./")
288+
et = ExtraTreesRegressor()
289+
et.fit(X_train, y_train)
290+
y_pred = et.predict(X_test)
291+
print("flamlized et r2", "=", 1 - sklearn_metric_loss_score("r2", y_pred, y_test))
292+
print(et)
293+
```
294+
295+
#### Sample output
296+
297+
```
298+
load dataset from ./openml_ds537.pkl
299+
Dataset name: houses
300+
X_train.shape: (15480, 8), y_train.shape: (15480,);
301+
X_test.shape: (5160, 8), y_test.shape: (5160,)
302+
flamlized et r2 = 0.8534
303+
ExtraTreesRegressor(max_features=0.75, min_samples_leaf=2, min_samples_split=5,
304+
n_estimators=500)
305+
```
306+
307+
## Flamlized ExtraTreesClassifier
308+
309+
### Prerequisites
310+
311+
This example requires the [autozero] option.
312+
313+
```bash
314+
pip install flaml[autozero] scikit-learn openml
315+
```
316+
317+
### Zero-shot AutoML
318+
319+
```python
320+
from flaml.automl.data import load_openml_dataset
321+
from flaml.default import ExtraTreesClassifier
322+
from flaml.automl.ml import sklearn_metric_loss_score
323+
324+
X_train, X_test, y_train, y_test = load_openml_dataset(dataset_id=1169, data_dir="./")
325+
et = ExtraTreesClassifier()
326+
et.fit(X_train, y_train)
327+
y_pred = et.predict(X_test)
328+
print(
329+
"flamlized et accuracy",
330+
"=",
331+
1 - sklearn_metric_loss_score("accuracy", y_pred, y_test),
332+
)
333+
print(et)
334+
```
335+
336+
#### Sample output
337+
338+
```
339+
load dataset from ./openml_ds1169.pkl
340+
Dataset name: airlines
341+
X_train.shape: (404537, 7), y_train.shape: (404537,);
342+
X_test.shape: (134846, 7), y_test.shape: (134846,)
343+
flamlized et accuracy = 0.6698
344+
ExtraTreesClassifier(max_features=0.7, min_samples_leaf=3, min_samples_split=5,
345+
n_estimators=500)
346+
```

0 commit comments

Comments
 (0)