Skip to content

Commit f146815

Browse files
razdoburdinnapetrov
razdoburdin
and
napetrov
authored
Daal4py gbt support (#138)
* initial * pep8 * Replace np.iinfo(np.int32).max by 2**31 * Update configs/daal4py/daal4py_gbt_config.json Co-authored-by: Nikolay Petrov <nikolay.a.petrov@intel.com> * Update daal4py_bench/gbt.py Co-authored-by: Nikolay Petrov <nikolay.a.petrov@intel.com> --------- Co-authored-by: Dmitry Razdoburdin <> Co-authored-by: Nikolay Petrov <nikolay.a.petrov@intel.com>
1 parent 170fd4f commit f146815

File tree

2 files changed

+458
-0
lines changed

2 files changed

+458
-0
lines changed

‎configs/daal4py/daal4py_gbt_config.json

Lines changed: 280 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,280 @@
1+
{
2+
"common": {
3+
"lib": "daal4py",
4+
"data-format": "pandas",
5+
"data-order": "F",
6+
"fptype": "float",
7+
"algorithm": "gbt"
8+
},
9+
"cases": [
10+
{
11+
"dataset": [
12+
{
13+
"source": "npy",
14+
"name": "abalone",
15+
"training": {
16+
"x": "data/abalone_x_train.npy",
17+
"y": "data/abalone_y_train.npy"
18+
},
19+
"testing": {
20+
"x": "data/abalone_x_test.npy",
21+
"y": "data/abalone_y_test.npy"
22+
}
23+
}
24+
],
25+
"max_tree_depth": 6,
26+
"n_estimators": 1000,
27+
"objective": "reg:squarederror"
28+
},
29+
{
30+
"dataset": [
31+
{
32+
"source": "npy",
33+
"name": "airline-ohe",
34+
"training": {
35+
"x": "data/airline-ohe_x_train.npy",
36+
"y": "data/airline-ohe_y_train.npy"
37+
},
38+
"testing": {
39+
"x": "data/airline-ohe_x_test.npy",
40+
"y": "data/airline-ohe_y_test.npy"
41+
}
42+
}
43+
],
44+
"reg_lambda": 1,
45+
"max_tree_depth": 8,
46+
"n_estimators": 1000,
47+
"objective": "binary:logistic"
48+
},
49+
{
50+
"dataset": [
51+
{
52+
"source": "npy",
53+
"name": "higgs1m",
54+
"training": {
55+
"x": "data/higgs1m_x_train.npy",
56+
"y": "data/higgs1m_y_train.npy"
57+
},
58+
"testing": {
59+
"x": "data/higgs1m_x_test.npy",
60+
"y": "data/higgs1m_y_test.npy"
61+
}
62+
}
63+
],
64+
"max_bins": 256,
65+
"reg_lambda": 1,
66+
"max_tree_depth": 8,
67+
"n_estimators": [100, 300, 1000, 3000, 10000, 30000],
68+
"objective": "binary:logistic"
69+
},
70+
{
71+
"dataset": [
72+
{
73+
"source": "npy",
74+
"name": "letters",
75+
"training": {
76+
"x": "data/letters_x_train.npy",
77+
"y": "data/letters_y_train.npy"
78+
},
79+
"testing": {
80+
"x": "data/letters_x_test.npy",
81+
"y": "data/letters_y_test.npy"
82+
}
83+
}
84+
],
85+
"max_tree_depth": 6,
86+
"n_estimators": 1000,
87+
"objective": "multi:softprob"
88+
},
89+
{
90+
"dataset": [
91+
{
92+
"source": "npy",
93+
"name": "mlsr",
94+
"training": {
95+
"x": "data/mlsr_x_train.npy",
96+
"y": "data/mlsr_y_train.npy"
97+
}
98+
}
99+
],
100+
"max_bins": 256,
101+
"reg_lambda": 2,
102+
"max_tree_depth": 8,
103+
"n_estimators": 200,
104+
"objective": "multi:softprob"
105+
},
106+
{
107+
"dataset": [
108+
{
109+
"source": "npy",
110+
"name": "mortgage1Q",
111+
"training": {
112+
"x": "data/mortgage1Q_x_train.npy",
113+
"y": "data/mortgage1Q_y_train.npy"
114+
}
115+
}
116+
],
117+
"n_estimators": 100,
118+
"objective": "reg:squarederror",
119+
"max_tree_depth": 8,
120+
"reg_lambda": 1
121+
},
122+
{
123+
"dataset": [
124+
{
125+
"source": "npy",
126+
"name": "plasticc",
127+
"training": {
128+
"x": "data/plasticc_x_train.npy",
129+
"y": "data/plasticc_y_train.npy"
130+
},
131+
"testing": {
132+
"x": "data/plasticc_x_test.npy",
133+
"y": "data/plasticc_y_test.npy"
134+
}
135+
}
136+
],
137+
"n_estimators": 60,
138+
"objective": "multi:softprob",
139+
"max_tree_depth": 7
140+
},
141+
{
142+
"dataset": [
143+
{
144+
"source": "npy",
145+
"name": "santander",
146+
"training": {
147+
"x": "data/santander_x_train.npy",
148+
"y": "data/santander_y_train.npy"
149+
},
150+
"testing": {
151+
"x": "data/santander_x_test.npy",
152+
"y": "data/santander_y_test.npy"
153+
}
154+
}
155+
],
156+
"n_estimators": 10000,
157+
"objective": "binary:logistic",
158+
"max_tree_depth": 1
159+
},
160+
{
161+
"objective": "binary:logistic",
162+
"dataset": [
163+
{
164+
"source": "npy",
165+
"name": "airline",
166+
"training": {
167+
"x": "data/airline_x_train.npy",
168+
"y": "data/airline_y_train.npy"
169+
},
170+
"testing": {
171+
"x": "data/airline_x_test.npy",
172+
"y": "data/airline_y_test.npy"
173+
}
174+
}
175+
]
176+
},
177+
{
178+
"objective": "binary:logistic",
179+
"dataset": [
180+
{
181+
"source": "npy",
182+
"name": "bosch",
183+
"training": {
184+
"x": "data/bosch_x_train.npy",
185+
"y": "data/bosch_y_train.npy"
186+
},
187+
"testing": {
188+
"x": "data/bosch_x_test.npy",
189+
"y": "data/bosch_y_test.npy"
190+
}
191+
}
192+
]
193+
},
194+
{
195+
"objective": "multi:softmax",
196+
"dataset": [
197+
{
198+
"source": "npy",
199+
"name": "covtype",
200+
"training": {
201+
"x": "data/covtype_x_train.npy",
202+
"y": "data/covtype_y_train.npy"
203+
},
204+
"testing": {
205+
"x": "data/covtype_x_test.npy",
206+
"y": "data/covtype_y_test.npy"
207+
}
208+
}
209+
]
210+
},
211+
{
212+
"objective": "binary:logistic",
213+
"dataset": [
214+
{
215+
"source": "npy",
216+
"name": "epsilon",
217+
"training": {
218+
"x": "data/epsilon_x_train.npy",
219+
"y": "data/epsilon_y_train.npy"
220+
},
221+
"testing": {
222+
"x": "data/epsilon_x_test.npy",
223+
"y": "data/epsilon_y_test.npy"
224+
}
225+
}
226+
]
227+
},
228+
{
229+
"objective": "binary:logistic",
230+
"dataset": [
231+
{
232+
"source": "npy",
233+
"name": "fraud",
234+
"training": {
235+
"x": "data/fraud_x_train.npy",
236+
"y": "data/fraud_y_train.npy"
237+
},
238+
"testing": {
239+
"x": "data/fraud_x_test.npy",
240+
"y": "data/fraud_y_test.npy"
241+
}
242+
}
243+
]
244+
},
245+
{
246+
"objective": "binary:logistic",
247+
"dataset": [
248+
{
249+
"source": "npy",
250+
"name": "higgs",
251+
"training": {
252+
"x": "data/higgs_x_train.npy",
253+
"y": "data/higgs_y_train.npy"
254+
},
255+
"testing": {
256+
"x": "data/higgs_x_test.npy",
257+
"y": "data/higgs_y_test.npy"
258+
}
259+
}
260+
]
261+
},
262+
{
263+
"objective": "reg:squarederror",
264+
"dataset": [
265+
{
266+
"source": "npy",
267+
"name": "year_prediction_msd",
268+
"training": {
269+
"x": "data/year_prediction_msd_x_train.npy",
270+
"y": "data/year_prediction_msd_y_train.npy"
271+
},
272+
"testing": {
273+
"x": "data/year_prediction_msd_x_test.npy",
274+
"y": "data/year_prediction_msd_y_test.npy"
275+
}
276+
}
277+
]
278+
}
279+
]
280+
}

0 commit comments

Comments
 (0)