|
a |
|
b/GradientBoosting (1).ipynb |
|
|
1 |
{ |
|
|
2 |
"cells": [ |
|
|
3 |
{ |
|
|
4 |
"cell_type": "code", |
|
|
5 |
"execution_count": 99, |
|
|
6 |
"metadata": { |
|
|
7 |
"colab": { |
|
|
8 |
"base_uri": "https://localhost:8080/" |
|
|
9 |
}, |
|
|
10 |
"id": "9eWjYbUGEr1O", |
|
|
11 |
"outputId": "85ddb237-5139-454a-c9fd-5323f47aec3f" |
|
|
12 |
}, |
|
|
13 |
"outputs": [ |
|
|
14 |
{ |
|
|
15 |
"output_type": "stream", |
|
|
16 |
"name": "stdout", |
|
|
17 |
"text": [ |
|
|
18 |
"Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n" |
|
|
19 |
] |
|
|
20 |
} |
|
|
21 |
], |
|
|
22 |
"source": [ |
|
|
23 |
"from google.colab import drive\n", |
|
|
24 |
"drive.mount('/content/drive')" |
|
|
25 |
], |
|
|
26 |
"id": "9eWjYbUGEr1O" |
|
|
27 |
}, |
|
|
28 |
{ |
|
|
29 |
"cell_type": "code", |
|
|
30 |
"execution_count": 100, |
|
|
31 |
"metadata": { |
|
|
32 |
"id": "6673470e-db27-403d-8830-b4477fc8b6ca" |
|
|
33 |
}, |
|
|
34 |
"outputs": [], |
|
|
35 |
"source": [ |
|
|
36 |
"import pandas as pd\n", |
|
|
37 |
"import numpy as np\n", |
|
|
38 |
"import seaborn as sns\n", |
|
|
39 |
"import matplotlib.pyplot as plt\n", |
|
|
40 |
"from imblearn.over_sampling import SMOTE\n", |
|
|
41 |
"from imblearn.under_sampling import NearMiss\n", |
|
|
42 |
"from sklearn.model_selection import train_test_split, RandomizedSearchCV\n", |
|
|
43 |
"from sklearn.preprocessing import StandardScaler\n", |
|
|
44 |
"from sklearn.ensemble import GradientBoostingClassifier\n", |
|
|
45 |
"from imblearn.pipeline import Pipeline as ImbPipeline\n", |
|
|
46 |
"from sklearn.metrics import classification_report\n", |
|
|
47 |
"from sklearn.utils.class_weight import compute_sample_weight\n", |
|
|
48 |
"from collections import Counter\n", |
|
|
49 |
"import warnings\n", |
|
|
50 |
"warnings.filterwarnings('ignore')\n", |
|
|
51 |
"warnings.filterwarnings(\"ignore\", category=UserWarning, module=\"joblib\")\n", |
|
|
52 |
"warnings.filterwarnings(\"ignore\", category=UserWarning, module=\"sklearn\")\n", |
|
|
53 |
"warnings.filterwarnings('ignore', category=UserWarning, message=\"Line Search failed\")" |
|
|
54 |
], |
|
|
55 |
"id": "6673470e-db27-403d-8830-b4477fc8b6ca" |
|
|
56 |
}, |
|
|
57 |
{ |
|
|
58 |
"cell_type": "code", |
|
|
59 |
"execution_count": 101, |
|
|
60 |
"metadata": { |
|
|
61 |
"colab": { |
|
|
62 |
"base_uri": "https://localhost:8080/" |
|
|
63 |
}, |
|
|
64 |
"id": "21e2199b-7ab7-42cf-8664-150a744bbaae", |
|
|
65 |
"outputId": "ec0558c1-3eff-4923-8f60-3d360d66260e" |
|
|
66 |
}, |
|
|
67 |
"outputs": [ |
|
|
68 |
{ |
|
|
69 |
"output_type": "stream", |
|
|
70 |
"name": "stdout", |
|
|
71 |
"text": [ |
|
|
72 |
" Name FC logFC logCPM P-Value FDR SCLC NSCLC\n", |
|
|
73 |
"0 KRT16P5 -1.474275 -0.560006 -2.065784 0.423250 0.645529 0.0 0.0\n", |
|
|
74 |
"1 KRT16P3 -1.158475 -0.212227 0.698547 0.598622 0.779482 0.0 0.0\n", |
|
|
75 |
"2 KRT16P2 1.785481 0.836313 3.744968 0.060200 0.211667 0.0 0.0\n", |
|
|
76 |
"3 KRT16P6 -2.534136 -1.341494 0.404997 0.023716 0.123727 0.0 0.0\n", |
|
|
77 |
"4 CRHBP 1.441891 0.527962 -0.015277 0.034942 0.153404 0.0 0.0\n", |
|
|
78 |
"Name 0\n", |
|
|
79 |
"FC 0\n", |
|
|
80 |
"logFC 0\n", |
|
|
81 |
"logCPM 0\n", |
|
|
82 |
"P-Value 0\n", |
|
|
83 |
"FDR 0\n", |
|
|
84 |
"SCLC 0\n", |
|
|
85 |
"NSCLC 0\n", |
|
|
86 |
"dtype: int64\n" |
|
|
87 |
] |
|
|
88 |
} |
|
|
89 |
], |
|
|
90 |
"source": [ |
|
|
91 |
"#load data and preprocess\n", |
|
|
92 |
"file_path = \"/content/drive/MyDrive/ML_HW_4_5/labelled data.csv\"\n", |
|
|
93 |
"data = pd.read_csv(file_path,index_col=0).fillna(0)\n", |
|
|
94 |
"print(data.head())\n", |
|
|
95 |
"print(data.isnull().sum())" |
|
|
96 |
], |
|
|
97 |
"id": "21e2199b-7ab7-42cf-8664-150a744bbaae" |
|
|
98 |
}, |
|
|
99 |
{ |
|
|
100 |
"cell_type": "code", |
|
|
101 |
"source": [ |
|
|
102 |
"print(data.shape)\n", |
|
|
103 |
"print()\n", |
|
|
104 |
"print(data.describe)\n", |
|
|
105 |
"print()\n", |
|
|
106 |
"print(data.info)\n", |
|
|
107 |
"print()\n", |
|
|
108 |
"print(data.duplicated())\n", |
|
|
109 |
"print()\n", |
|
|
110 |
"print(data.dtypes)" |
|
|
111 |
], |
|
|
112 |
"metadata": { |
|
|
113 |
"colab": { |
|
|
114 |
"base_uri": "https://localhost:8080/" |
|
|
115 |
}, |
|
|
116 |
"id": "kdbbvktnjXR9", |
|
|
117 |
"outputId": "c299763e-4e15-49a9-afe8-6f0b85551152" |
|
|
118 |
}, |
|
|
119 |
"id": "kdbbvktnjXR9", |
|
|
120 |
"execution_count": 102, |
|
|
121 |
"outputs": [ |
|
|
122 |
{ |
|
|
123 |
"output_type": "stream", |
|
|
124 |
"name": "stdout", |
|
|
125 |
"text": [ |
|
|
126 |
"(19778, 8)\n", |
|
|
127 |
"\n", |
|
|
128 |
"<bound method NDFrame.describe of Name FC logFC logCPM P-Value FDR SCLC \\\n", |
|
|
129 |
"0 KRT16P5 -1.474275 -0.560006 -2.065784 0.423250 0.645529 0.0 \n", |
|
|
130 |
"1 KRT16P3 -1.158475 -0.212227 0.698547 0.598622 0.779482 0.0 \n", |
|
|
131 |
"2 KRT16P2 1.785481 0.836313 3.744968 0.060200 0.211667 0.0 \n", |
|
|
132 |
"3 KRT16P6 -2.534136 -1.341494 0.404997 0.023716 0.123727 0.0 \n", |
|
|
133 |
"4 CRHBP 1.441891 0.527962 -0.015277 0.034942 0.153404 0.0 \n", |
|
|
134 |
"... ... ... ... ... ... ... ... \n", |
|
|
135 |
"19773 LOC105369958 2.196994 1.135531 1.382694 0.003199 0.040235 0.0 \n", |
|
|
136 |
"19774 ABCC2 3.461301 1.791314 5.336636 0.000671 0.015864 0.0 \n", |
|
|
137 |
"19775 TRAV6 3.849574 1.944699 -0.401490 0.000016 0.001372 0.0 \n", |
|
|
138 |
"19776 LOC105369904 2.226049 1.154485 -1.006838 0.002616 0.035713 0.0 \n", |
|
|
139 |
"19777 LOC101928636 2.279563 1.188757 0.151617 0.000219 0.007721 0.0 \n", |
|
|
140 |
"\n", |
|
|
141 |
" NSCLC \n", |
|
|
142 |
"0 0.0 \n", |
|
|
143 |
"1 0.0 \n", |
|
|
144 |
"2 0.0 \n", |
|
|
145 |
"3 0.0 \n", |
|
|
146 |
"4 0.0 \n", |
|
|
147 |
"... ... \n", |
|
|
148 |
"19773 0.0 \n", |
|
|
149 |
"19774 0.0 \n", |
|
|
150 |
"19775 0.0 \n", |
|
|
151 |
"19776 0.0 \n", |
|
|
152 |
"19777 0.0 \n", |
|
|
153 |
"\n", |
|
|
154 |
"[19778 rows x 8 columns]>\n", |
|
|
155 |
"\n", |
|
|
156 |
"<bound method DataFrame.info of Name FC logFC logCPM P-Value FDR SCLC \\\n", |
|
|
157 |
"0 KRT16P5 -1.474275 -0.560006 -2.065784 0.423250 0.645529 0.0 \n", |
|
|
158 |
"1 KRT16P3 -1.158475 -0.212227 0.698547 0.598622 0.779482 0.0 \n", |
|
|
159 |
"2 KRT16P2 1.785481 0.836313 3.744968 0.060200 0.211667 0.0 \n", |
|
|
160 |
"3 KRT16P6 -2.534136 -1.341494 0.404997 0.023716 0.123727 0.0 \n", |
|
|
161 |
"4 CRHBP 1.441891 0.527962 -0.015277 0.034942 0.153404 0.0 \n", |
|
|
162 |
"... ... ... ... ... ... ... ... \n", |
|
|
163 |
"19773 LOC105369958 2.196994 1.135531 1.382694 0.003199 0.040235 0.0 \n", |
|
|
164 |
"19774 ABCC2 3.461301 1.791314 5.336636 0.000671 0.015864 0.0 \n", |
|
|
165 |
"19775 TRAV6 3.849574 1.944699 -0.401490 0.000016 0.001372 0.0 \n", |
|
|
166 |
"19776 LOC105369904 2.226049 1.154485 -1.006838 0.002616 0.035713 0.0 \n", |
|
|
167 |
"19777 LOC101928636 2.279563 1.188757 0.151617 0.000219 0.007721 0.0 \n", |
|
|
168 |
"\n", |
|
|
169 |
" NSCLC \n", |
|
|
170 |
"0 0.0 \n", |
|
|
171 |
"1 0.0 \n", |
|
|
172 |
"2 0.0 \n", |
|
|
173 |
"3 0.0 \n", |
|
|
174 |
"4 0.0 \n", |
|
|
175 |
"... ... \n", |
|
|
176 |
"19773 0.0 \n", |
|
|
177 |
"19774 0.0 \n", |
|
|
178 |
"19775 0.0 \n", |
|
|
179 |
"19776 0.0 \n", |
|
|
180 |
"19777 0.0 \n", |
|
|
181 |
"\n", |
|
|
182 |
"[19778 rows x 8 columns]>\n", |
|
|
183 |
"\n", |
|
|
184 |
"0 False\n", |
|
|
185 |
"1 False\n", |
|
|
186 |
"2 False\n", |
|
|
187 |
"3 False\n", |
|
|
188 |
"4 False\n", |
|
|
189 |
" ... \n", |
|
|
190 |
"19773 False\n", |
|
|
191 |
"19774 False\n", |
|
|
192 |
"19775 False\n", |
|
|
193 |
"19776 False\n", |
|
|
194 |
"19777 False\n", |
|
|
195 |
"Length: 19778, dtype: bool\n", |
|
|
196 |
"\n", |
|
|
197 |
"Name object\n", |
|
|
198 |
"FC float64\n", |
|
|
199 |
"logFC float64\n", |
|
|
200 |
"logCPM float64\n", |
|
|
201 |
"P-Value float64\n", |
|
|
202 |
"FDR float64\n", |
|
|
203 |
"SCLC float64\n", |
|
|
204 |
"NSCLC float64\n", |
|
|
205 |
"dtype: object\n" |
|
|
206 |
] |
|
|
207 |
} |
|
|
208 |
] |
|
|
209 |
}, |
|
|
210 |
{ |
|
|
211 |
"cell_type": "code", |
|
|
212 |
"source": [ |
|
|
213 |
"print(data.columns[data.isna().any()])\n", |
|
|
214 |
"print()" |
|
|
215 |
], |
|
|
216 |
"metadata": { |
|
|
217 |
"colab": { |
|
|
218 |
"base_uri": "https://localhost:8080/" |
|
|
219 |
}, |
|
|
220 |
"id": "njDFuOYEjXBr", |
|
|
221 |
"outputId": "b0afc5e1-2326-46c0-9d25-5ccfcfa6580f" |
|
|
222 |
}, |
|
|
223 |
"id": "njDFuOYEjXBr", |
|
|
224 |
"execution_count": 103, |
|
|
225 |
"outputs": [ |
|
|
226 |
{ |
|
|
227 |
"output_type": "stream", |
|
|
228 |
"name": "stdout", |
|
|
229 |
"text": [ |
|
|
230 |
"Index([], dtype='object')\n", |
|
|
231 |
"\n" |
|
|
232 |
] |
|
|
233 |
} |
|
|
234 |
] |
|
|
235 |
}, |
|
|
236 |
{ |
|
|
237 |
"cell_type": "code", |
|
|
238 |
"source": [ |
|
|
239 |
"correlation_matrix = data.corr()\n", |
|
|
240 |
"# Create a heatmap using Seaborn\n", |
|
|
241 |
"plt.figure(figsize=(10, 8))\n", |
|
|
242 |
"sns.heatmap(correlation_matrix, annot=True, cmap=\"coolwarm\", fmt=\".2f\", linewidths=.5)\n", |
|
|
243 |
"plt.title(\"Correlation Matrix\")\n", |
|
|
244 |
"plt.show()" |
|
|
245 |
], |
|
|
246 |
"metadata": { |
|
|
247 |
"colab": { |
|
|
248 |
"base_uri": "https://localhost:8080/", |
|
|
249 |
"height": 699 |
|
|
250 |
}, |
|
|
251 |
"id": "Eo1IltL7X_l5", |
|
|
252 |
"outputId": "7ada2bf3-1c8f-4546-d479-40dcfdf53048" |
|
|
253 |
}, |
|
|
254 |
"id": "Eo1IltL7X_l5", |
|
|
255 |
"execution_count": 89, |
|
|
256 |
"outputs": [ |
|
|
257 |
{ |
|
|
258 |
"output_type": "display_data", |
|
|
259 |
"data": { |
|
|
260 |
"text/plain": [ |
|
|
261 |
"<Figure size 1000x800 with 2 Axes>" |
|
|
262 |
], |
|
|
263 |
"image/png": "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\n" |
|
|
264 |
}, |
|
|
265 |
"metadata": {} |
|
|
266 |
} |
|
|
267 |
] |
|
|
268 |
}, |
|
|
269 |
{ |
|
|
270 |
"cell_type": "code", |
|
|
271 |
"execution_count": 90, |
|
|
272 |
"metadata": { |
|
|
273 |
"id": "727e6842-3bc5-4e52-a7d2-fb2fc513d41c" |
|
|
274 |
}, |
|
|
275 |
"outputs": [], |
|
|
276 |
"source": [ |
|
|
277 |
"#feature selection\n", |
|
|
278 |
"features = data[['FC', 'logFC', 'P-Value']]\n", |
|
|
279 |
"targets = {'NSCLC': data['NSCLC'], 'SCLC': data['SCLC']}" |
|
|
280 |
], |
|
|
281 |
"id": "727e6842-3bc5-4e52-a7d2-fb2fc513d41c" |
|
|
282 |
}, |
|
|
283 |
{ |
|
|
284 |
"cell_type": "code", |
|
|
285 |
"source": [ |
|
|
286 |
"nsclc = data['NSCLC']\n", |
|
|
287 |
"sclc = data['SCLC']" |
|
|
288 |
], |
|
|
289 |
"metadata": { |
|
|
290 |
"id": "Qkr6jiCJP2r0" |
|
|
291 |
}, |
|
|
292 |
"id": "Qkr6jiCJP2r0", |
|
|
293 |
"execution_count": 91, |
|
|
294 |
"outputs": [] |
|
|
295 |
}, |
|
|
296 |
{ |
|
|
297 |
"cell_type": "code", |
|
|
298 |
"source": [ |
|
|
299 |
"nm = NearMiss()\n", |
|
|
300 |
"print('SCLC Original Shape:', Counter(sclc))\n", |
|
|
301 |
"features_nm_sclc, nm_sclc = nm.fit_resample(features, sclc)\n", |
|
|
302 |
"print('SCLC Resample Shape:', Counter(nm_sclc))\n", |
|
|
303 |
"print('NSCLC Original Shape:', Counter(nsclc))\n", |
|
|
304 |
"features_nm_nsclc, nm_nsclc = nm.fit_resample(features, nsclc)\n", |
|
|
305 |
"print('NSCLC Resample Shape:', Counter(nm_nsclc))" |
|
|
306 |
], |
|
|
307 |
"metadata": { |
|
|
308 |
"colab": { |
|
|
309 |
"base_uri": "https://localhost:8080/" |
|
|
310 |
}, |
|
|
311 |
"id": "hUFqQREwP241", |
|
|
312 |
"outputId": "0ab5204f-2b2a-4eff-a85e-69b652ed989c" |
|
|
313 |
}, |
|
|
314 |
"id": "hUFqQREwP241", |
|
|
315 |
"execution_count": 92, |
|
|
316 |
"outputs": [ |
|
|
317 |
{ |
|
|
318 |
"output_type": "stream", |
|
|
319 |
"name": "stdout", |
|
|
320 |
"text": [ |
|
|
321 |
"SCLC Original Shape: Counter({0.0: 18857, 1.0: 921})\n", |
|
|
322 |
"SCLC Resample Shape: Counter({0.0: 921, 1.0: 921})\n", |
|
|
323 |
"NSCLC Original Shape: Counter({0.0: 19087, 1.0: 691})\n", |
|
|
324 |
"NSCLC Resample Shape: Counter({0.0: 691, 1.0: 691})\n" |
|
|
325 |
] |
|
|
326 |
} |
|
|
327 |
] |
|
|
328 |
}, |
|
|
329 |
{ |
|
|
330 |
"cell_type": "code", |
|
|
331 |
"source": [ |
|
|
332 |
"features_sclc = features_nm_sclc\n", |
|
|
333 |
"features_nsclc = features_nm_nsclc\n", |
|
|
334 |
"sclc = nm_sclc\n", |
|
|
335 |
"nsclc = nm_nsclc" |
|
|
336 |
], |
|
|
337 |
"metadata": { |
|
|
338 |
"id": "2_FJdkOjQCL5" |
|
|
339 |
}, |
|
|
340 |
"id": "2_FJdkOjQCL5", |
|
|
341 |
"execution_count": 93, |
|
|
342 |
"outputs": [] |
|
|
343 |
}, |
|
|
344 |
{ |
|
|
345 |
"cell_type": "code", |
|
|
346 |
"execution_count": 94, |
|
|
347 |
"metadata": { |
|
|
348 |
"colab": { |
|
|
349 |
"base_uri": "https://localhost:8080/", |
|
|
350 |
"height": 963 |
|
|
351 |
}, |
|
|
352 |
"id": "be436149-9bb1-4c02-8373-2fd80c97f05c", |
|
|
353 |
"outputId": "98ebf4c9-fbef-44d6-d8e3-8de05fed158b" |
|
|
354 |
}, |
|
|
355 |
"outputs": [ |
|
|
356 |
{ |
|
|
357 |
"output_type": "display_data", |
|
|
358 |
"data": { |
|
|
359 |
"text/plain": [ |
|
|
360 |
"<Figure size 1500x600 with 4 Axes>" |
|
|
361 |
], |
|
|
362 |
"image/png": "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\n" |
|
|
363 |
}, |
|
|
364 |
"metadata": {} |
|
|
365 |
}, |
|
|
366 |
{ |
|
|
367 |
"output_type": "display_data", |
|
|
368 |
"data": { |
|
|
369 |
"text/plain": [ |
|
|
370 |
"<Figure size 640x480 with 2 Axes>" |
|
|
371 |
], |
|
|
372 |
"image/png": "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\n" |
|
|
373 |
}, |
|
|
374 |
"metadata": {} |
|
|
375 |
} |
|
|
376 |
], |
|
|
377 |
"source": [ |
|
|
378 |
"#EDA\n", |
|
|
379 |
"#histograms and Correlation Matrix\n", |
|
|
380 |
"features.hist(bins=15, figsize=(15, 6), layout=(2, 2))\n", |
|
|
381 |
"plt.show()\n", |
|
|
382 |
"sns.heatmap(features.corr(), annot=True)\n", |
|
|
383 |
"plt.show()" |
|
|
384 |
], |
|
|
385 |
"id": "be436149-9bb1-4c02-8373-2fd80c97f05c" |
|
|
386 |
}, |
|
|
387 |
{ |
|
|
388 |
"cell_type": "code", |
|
|
389 |
"execution_count": 95, |
|
|
390 |
"metadata": { |
|
|
391 |
"id": "1bef77ce-9eb7-444d-963f-47d25ea757c1" |
|
|
392 |
}, |
|
|
393 |
"outputs": [], |
|
|
394 |
"source": [ |
|
|
395 |
"def train_test_and_standardize(features, target, test_size=0.2, random_state=42):\n", |
|
|
396 |
" # Split the data into train and test sets\n", |
|
|
397 |
" x_train, x_test, y_train, y_test = train_test_split(features, target, test_size=test_size, random_state=random_state)\n", |
|
|
398 |
"\n", |
|
|
399 |
" # Standardize the features using StandardScaler\n", |
|
|
400 |
" scaler = StandardScaler()\n", |
|
|
401 |
" x_train = scaler.fit_transform(x_train)\n", |
|
|
402 |
" x_test = scaler.transform(x_test)\n", |
|
|
403 |
"\n", |
|
|
404 |
" return x_train, x_test, y_train, y_test\n", |
|
|
405 |
"\n", |
|
|
406 |
"xtrain_sclc, xtest_sclc, ytrain_sclc, ytest_sclc = train_test_and_standardize(features_sclc, sclc)\n", |
|
|
407 |
"xtrain_nsclc, xtest_nsclc, ytrain_nsclc, ytest_nsclc = train_test_and_standardize(features_nsclc, nsclc)" |
|
|
408 |
], |
|
|
409 |
"id": "1bef77ce-9eb7-444d-963f-47d25ea757c1" |
|
|
410 |
}, |
|
|
411 |
{ |
|
|
412 |
"cell_type": "code", |
|
|
413 |
"execution_count": 96, |
|
|
414 |
"metadata": { |
|
|
415 |
"id": "27fbdfa2-ccfb-4ac4-b09f-e1082e45294c" |
|
|
416 |
}, |
|
|
417 |
"outputs": [], |
|
|
418 |
"source": [ |
|
|
419 |
"# Update your pipeline with class_weight parameter\n", |
|
|
420 |
"pipeline = ImbPipeline([\n", |
|
|
421 |
" ('scaler', StandardScaler()),\n", |
|
|
422 |
" ('gradient_boosting', GradientBoostingClassifier())\n", |
|
|
423 |
"])\n", |
|
|
424 |
"\n", |
|
|
425 |
"# Function to fit the pipeline with class weights\n", |
|
|
426 |
"def fit_pipeline_with_weights(pipeline, x_train, y_train, class_weights):\n", |
|
|
427 |
" pipeline.named_steps['gradient_boosting'].sample_weight = class_weights\n", |
|
|
428 |
" pipeline.fit(x_train, y_train)\n", |
|
|
429 |
" return pipeline\n", |
|
|
430 |
"\n", |
|
|
431 |
"# Update your parameter grid to include more hyperparameters or adjust the ranges\n", |
|
|
432 |
"param_grid_gradient_boosting = {\n", |
|
|
433 |
" 'gradient_boosting__loss': ['deviance', 'exponential'],\n", |
|
|
434 |
" 'gradient_boosting__learning_rate': [0.01, 0.1, 0.2], # Adjust the learning rate\n", |
|
|
435 |
" 'gradient_boosting__n_estimators': [200, 300, 400], # Increase the number of estimators\n", |
|
|
436 |
" 'gradient_boosting__subsample': [1.0, 0.8, 0.6],\n", |
|
|
437 |
" 'gradient_boosting__criterion': ['friedman_mse', 'squared_error'],\n", |
|
|
438 |
" 'gradient_boosting__min_samples_split': [2, 4, 8],\n", |
|
|
439 |
" 'gradient_boosting__min_samples_leaf': [1, 2, 4],\n", |
|
|
440 |
" 'gradient_boosting__min_weight_fraction_leaf': [0.0, 0.1, 0.2],\n", |
|
|
441 |
" 'gradient_boosting__max_depth': [3, 5, 7],\n", |
|
|
442 |
" 'gradient_boosting__min_impurity_decrease': [0.0, 0.1, 0.2],\n", |
|
|
443 |
" 'gradient_boosting__init': [None, 'zero'],\n", |
|
|
444 |
" 'gradient_boosting__random_state': [None, 42, 100],\n", |
|
|
445 |
" 'gradient_boosting__max_features': [None, 'sqrt', 'log2'],\n", |
|
|
446 |
" 'gradient_boosting__verbose': [0, 1, 2],\n", |
|
|
447 |
" 'gradient_boosting__max_leaf_nodes': [None, 5, 10],\n", |
|
|
448 |
" 'gradient_boosting__warm_start': [False, True],\n", |
|
|
449 |
" 'gradient_boosting__validation_fraction': [0.1, 0.2, 0.3],\n", |
|
|
450 |
" 'gradient_boosting__n_iter_no_change': [None, 10, 20],\n", |
|
|
451 |
" 'gradient_boosting__tol': [1e-4, 1e-3, 1e-2],\n", |
|
|
452 |
" 'gradient_boosting__ccp_alpha': [0.0, 0.1, 0.2]\n", |
|
|
453 |
"}\n", |
|
|
454 |
"\n", |
|
|
455 |
"\n", |
|
|
456 |
"#function for GridSearchCV and model evaluation\n", |
|
|
457 |
"def evaluate_gradient_boosting(x_train, y_train, x_test, y_test, param_grid):\n", |
|
|
458 |
" random_search = RandomizedSearchCV(estimator=pipeline, param_distributions=param_grid, scoring='f1',n_iter = 1000, cv=10, verbose=1, n_jobs=-1)\n", |
|
|
459 |
" random_search.fit(x_train, y_train) #fit on training data\n", |
|
|
460 |
" best_params = random_search.best_params_\n", |
|
|
461 |
" best_score = random_search.best_score_\n", |
|
|
462 |
" best_gradient_boosting = pipeline.set_params(**best_params)\n", |
|
|
463 |
" best_gradient_boosting.fit(x_train, y_train) #refit on training data\n", |
|
|
464 |
" y_test_pred = best_gradient_boosting.predict(x_test) #predict on validation data\n", |
|
|
465 |
" report = classification_report(y_test, y_test_pred)\n", |
|
|
466 |
" return best_params, best_score, report" |
|
|
467 |
], |
|
|
468 |
"id": "27fbdfa2-ccfb-4ac4-b09f-e1082e45294c" |
|
|
469 |
}, |
|
|
470 |
{ |
|
|
471 |
"cell_type": "code", |
|
|
472 |
"execution_count": 97, |
|
|
473 |
"metadata": { |
|
|
474 |
"colab": { |
|
|
475 |
"base_uri": "https://localhost:8080/" |
|
|
476 |
}, |
|
|
477 |
"id": "28d9744d-ee67-4ba5-aa2b-d1ec35c0807f", |
|
|
478 |
"outputId": "b6dc5dd3-7dc7-42e8-a566-bda7963b661c" |
|
|
479 |
}, |
|
|
480 |
"outputs": [ |
|
|
481 |
{ |
|
|
482 |
"output_type": "stream", |
|
|
483 |
"name": "stdout", |
|
|
484 |
"text": [ |
|
|
485 |
"Fitting 10 folds for each of 1000 candidates, totalling 10000 fits\n", |
|
|
486 |
" Iter Train Loss OOB Improve Remaining Time \n", |
|
|
487 |
" 1 0.9954 0.0040 0.32s\n", |
|
|
488 |
" 2 0.9911 0.0048 0.40s\n", |
|
|
489 |
" 3 0.9871 0.0043 0.41s\n", |
|
|
490 |
" 4 0.9820 0.0036 0.42s\n", |
|
|
491 |
" 5 0.9785 0.0042 0.41s\n", |
|
|
492 |
" 6 0.9741 0.0041 0.41s\n", |
|
|
493 |
" 7 0.9698 0.0041 0.41s\n", |
|
|
494 |
" 8 0.9658 0.0037 0.41s\n", |
|
|
495 |
" 9 0.9614 0.0038 0.41s\n", |
|
|
496 |
" 10 0.9581 0.0040 0.41s\n", |
|
|
497 |
" 11 0.9525 0.0035 0.46s\n", |
|
|
498 |
" 12 0.9493 0.0042 0.46s\n", |
|
|
499 |
" 13 0.9445 0.0031 0.46s\n", |
|
|
500 |
" 14 0.9398 0.0037 0.45s\n", |
|
|
501 |
" 15 0.9366 0.0033 0.48s\n", |
|
|
502 |
" 16 0.9317 0.0033 0.48s\n", |
|
|
503 |
" 17 0.9289 0.0031 0.47s\n", |
|
|
504 |
" 18 0.9247 0.0035 0.47s\n", |
|
|
505 |
" 19 0.9230 0.0039 0.46s\n", |
|
|
506 |
" 20 0.9172 0.0033 0.46s\n", |
|
|
507 |
" 21 0.9154 0.0036 0.46s\n", |
|
|
508 |
" 22 0.9094 0.0027 0.45s\n", |
|
|
509 |
" 23 0.9106 0.0033 0.45s\n", |
|
|
510 |
" 24 0.9068 0.0038 0.45s\n", |
|
|
511 |
" 25 0.9009 0.0033 0.44s\n", |
|
|
512 |
" 26 0.8949 0.0032 0.44s\n", |
|
|
513 |
" 27 0.8942 0.0033 0.43s\n", |
|
|
514 |
" 28 0.8904 0.0031 0.43s\n", |
|
|
515 |
" 29 0.8887 0.0029 0.43s\n", |
|
|
516 |
" 30 0.8846 0.0029 0.43s\n", |
|
|
517 |
" 31 0.8787 0.0026 0.42s\n", |
|
|
518 |
" 32 0.8790 0.0028 0.42s\n", |
|
|
519 |
" 33 0.8764 0.0031 0.42s\n", |
|
|
520 |
" 34 0.8711 0.0030 0.41s\n", |
|
|
521 |
" 35 0.8676 0.0025 0.41s\n", |
|
|
522 |
" 36 0.8630 0.0027 0.41s\n", |
|
|
523 |
" 37 0.8622 0.0029 0.41s\n", |
|
|
524 |
" 38 0.8594 0.0026 0.41s\n", |
|
|
525 |
" 39 0.8531 0.0024 0.40s\n", |
|
|
526 |
" 40 0.8564 0.0031 0.40s\n", |
|
|
527 |
" 41 0.8448 0.0023 0.40s\n", |
|
|
528 |
" 42 0.8540 0.0033 0.40s\n", |
|
|
529 |
" 43 0.8461 0.0026 0.39s\n", |
|
|
530 |
" 44 0.8394 0.0023 0.39s\n", |
|
|
531 |
" 45 0.8393 0.0023 0.39s\n", |
|
|
532 |
" 46 0.8346 0.0020 0.39s\n", |
|
|
533 |
" 47 0.8359 0.0024 0.39s\n", |
|
|
534 |
" 48 0.8343 0.0027 0.38s\n", |
|
|
535 |
" 49 0.8240 0.0020 0.38s\n", |
|
|
536 |
" 50 0.8289 0.0022 0.38s\n", |
|
|
537 |
" 51 0.8209 0.0020 0.38s\n", |
|
|
538 |
" 52 0.8221 0.0024 0.38s\n", |
|
|
539 |
" 53 0.8189 0.0022 0.37s\n", |
|
|
540 |
" 54 0.8248 0.0029 0.37s\n", |
|
|
541 |
" 55 0.8113 0.0017 0.37s\n", |
|
|
542 |
" 56 0.8113 0.0019 0.37s\n", |
|
|
543 |
" 57 0.8052 0.0016 0.37s\n", |
|
|
544 |
" 58 0.8026 0.0017 0.37s\n", |
|
|
545 |
" 59 0.8027 0.0018 0.36s\n", |
|
|
546 |
" 60 0.8011 0.0018 0.36s\n", |
|
|
547 |
" 61 0.7980 0.0023 0.36s\n", |
|
|
548 |
" 62 0.7938 0.0021 0.36s\n", |
|
|
549 |
" 63 0.7976 0.0017 0.36s\n", |
|
|
550 |
" 64 0.7965 0.0024 0.36s\n", |
|
|
551 |
" 65 0.7948 0.0020 0.35s\n", |
|
|
552 |
" 66 0.7966 0.0021 0.35s\n", |
|
|
553 |
" 67 0.7842 0.0019 0.35s\n", |
|
|
554 |
" 68 0.7801 0.0019 0.35s\n", |
|
|
555 |
" 69 0.7747 0.0017 0.35s\n", |
|
|
556 |
" 70 0.7743 0.0018 0.35s\n", |
|
|
557 |
" 71 0.7786 0.0019 0.34s\n", |
|
|
558 |
" 72 0.7731 0.0021 0.34s\n", |
|
|
559 |
" 73 0.7814 0.0019 0.34s\n", |
|
|
560 |
" 74 0.7820 0.0021 0.34s\n", |
|
|
561 |
" 75 0.7609 0.0014 0.34s\n", |
|
|
562 |
" 76 0.7611 0.0015 0.33s\n", |
|
|
563 |
" 77 0.7674 0.0011 0.33s\n", |
|
|
564 |
" 78 0.7658 0.0016 0.33s\n", |
|
|
565 |
" 79 0.7664 0.0016 0.33s\n", |
|
|
566 |
" 80 0.7640 0.0019 0.33s\n", |
|
|
567 |
" 81 0.7581 0.0017 0.33s\n", |
|
|
568 |
" 82 0.7513 0.0009 0.33s\n", |
|
|
569 |
" 83 0.7522 0.0013 0.32s\n", |
|
|
570 |
" 84 0.7586 0.0016 0.32s\n", |
|
|
571 |
" 85 0.7490 0.0014 0.32s\n", |
|
|
572 |
" 86 0.7401 0.0011 0.32s\n", |
|
|
573 |
" 87 0.7410 0.0010 0.32s\n", |
|
|
574 |
" 88 0.7445 0.0015 0.31s\n", |
|
|
575 |
" 89 0.7398 0.0008 0.32s\n", |
|
|
576 |
" 90 0.7415 0.0010 0.31s\n", |
|
|
577 |
" 91 0.7467 0.0017 0.31s\n", |
|
|
578 |
" 92 0.7424 0.0014 0.31s\n", |
|
|
579 |
" 93 0.7374 0.0012 0.31s\n", |
|
|
580 |
" 94 0.7474 0.0014 0.31s\n", |
|
|
581 |
" 95 0.7366 0.0010 0.31s\n", |
|
|
582 |
" 96 0.7397 0.0014 0.30s\n", |
|
|
583 |
" 97 0.7325 0.0010 0.30s\n", |
|
|
584 |
" 98 0.7311 0.0009 0.30s\n", |
|
|
585 |
" 99 0.7350 0.0015 0.30s\n", |
|
|
586 |
" 100 0.7218 0.0007 0.30s\n", |
|
|
587 |
" 101 0.7172 0.0008 0.30s\n", |
|
|
588 |
" 102 0.7259 0.0014 0.29s\n", |
|
|
589 |
" 103 0.7236 0.0012 0.29s\n", |
|
|
590 |
" 104 0.7165 0.0008 0.29s\n", |
|
|
591 |
" 105 0.7236 0.0011 0.29s\n", |
|
|
592 |
" 106 0.7274 0.0012 0.29s\n", |
|
|
593 |
" 107 0.7200 0.0011 0.29s\n", |
|
|
594 |
" 108 0.7232 0.0009 0.28s\n", |
|
|
595 |
" 109 0.7197 0.0008 0.28s\n", |
|
|
596 |
" 110 0.7191 0.0012 0.28s\n", |
|
|
597 |
" 111 0.7189 0.0008 0.28s\n", |
|
|
598 |
" 112 0.7148 0.0008 0.28s\n", |
|
|
599 |
" 113 0.7095 0.0006 0.28s\n", |
|
|
600 |
" 114 0.7094 0.0007 0.28s\n", |
|
|
601 |
" 115 0.7096 0.0007 0.27s\n", |
|
|
602 |
" 116 0.7105 0.0010 0.27s\n", |
|
|
603 |
" 117 0.7001 0.0002 0.27s\n", |
|
|
604 |
" 118 0.7137 0.0010 0.27s\n", |
|
|
605 |
" 119 0.7095 0.0008 0.27s\n", |
|
|
606 |
" 120 0.6956 0.0008 0.27s\n", |
|
|
607 |
" 121 0.7000 0.0007 0.26s\n", |
|
|
608 |
" 122 0.7034 0.0006 0.26s\n", |
|
|
609 |
" 123 0.7033 0.0007 0.26s\n", |
|
|
610 |
" 124 0.7032 0.0008 0.26s\n", |
|
|
611 |
" 125 0.6959 0.0008 0.26s\n", |
|
|
612 |
" 126 0.6985 0.0006 0.26s\n", |
|
|
613 |
" 127 0.6811 0.0006 0.25s\n", |
|
|
614 |
" 128 0.6959 0.0005 0.25s\n", |
|
|
615 |
" 129 0.6947 0.0004 0.25s\n", |
|
|
616 |
" 130 0.7033 0.0012 0.25s\n", |
|
|
617 |
" 131 0.7061 0.0008 0.25s\n", |
|
|
618 |
" 132 0.6855 0.0006 0.25s\n", |
|
|
619 |
" 133 0.6920 0.0006 0.25s\n", |
|
|
620 |
" 134 0.6890 0.0005 0.24s\n", |
|
|
621 |
" 135 0.6812 0.0008 0.24s\n", |
|
|
622 |
" 136 0.6939 0.0009 0.24s\n", |
|
|
623 |
" 137 0.6912 0.0012 0.24s\n", |
|
|
624 |
" 138 0.6889 0.0004 0.24s\n", |
|
|
625 |
" 139 0.6892 0.0007 0.24s\n", |
|
|
626 |
" 140 0.6845 0.0010 0.24s\n", |
|
|
627 |
" 141 0.6806 0.0006 0.24s\n", |
|
|
628 |
" 142 0.6986 0.0005 0.23s\n", |
|
|
629 |
" 143 0.6684 0.0005 0.23s\n", |
|
|
630 |
" 144 0.6826 0.0008 0.23s\n", |
|
|
631 |
" 145 0.6711 0.0001 0.23s\n", |
|
|
632 |
" 146 0.6820 0.0010 0.23s\n", |
|
|
633 |
" 147 0.6850 0.0007 0.23s\n", |
|
|
634 |
" 148 0.6721 0.0007 0.22s\n", |
|
|
635 |
" 149 0.6625 0.0003 0.22s\n", |
|
|
636 |
" 150 0.6853 0.0008 0.22s\n", |
|
|
637 |
" 151 0.6715 0.0005 0.22s\n", |
|
|
638 |
" 152 0.6783 0.0005 0.22s\n", |
|
|
639 |
" 153 0.6714 0.0005 0.22s\n", |
|
|
640 |
" 154 0.6705 0.0005 0.22s\n", |
|
|
641 |
" 155 0.6849 0.0006 0.21s\n", |
|
|
642 |
" 156 0.6588 0.0003 0.21s\n", |
|
|
643 |
" 157 0.6761 0.0005 0.21s\n", |
|
|
644 |
" 158 0.6696 0.0004 0.21s\n", |
|
|
645 |
" 159 0.6665 0.0008 0.21s\n", |
|
|
646 |
" 160 0.6735 0.0005 0.21s\n", |
|
|
647 |
" 161 0.6624 0.0008 0.20s\n", |
|
|
648 |
" 162 0.6774 0.0005 0.20s\n", |
|
|
649 |
" 163 0.6715 0.0005 0.20s\n", |
|
|
650 |
" 164 0.6672 0.0006 0.20s\n", |
|
|
651 |
" 165 0.6601 0.0006 0.20s\n", |
|
|
652 |
" 166 0.6642 0.0004 0.20s\n", |
|
|
653 |
" 167 0.6696 0.0003 0.20s\n", |
|
|
654 |
" 168 0.6487 -0.0000 0.19s\n", |
|
|
655 |
" 169 0.6523 0.0005 0.19s\n", |
|
|
656 |
" 170 0.6512 0.0001 0.19s\n", |
|
|
657 |
" 171 0.6529 0.0005 0.19s\n", |
|
|
658 |
" 172 0.6588 0.0004 0.19s\n", |
|
|
659 |
" 173 0.6520 0.0004 0.19s\n", |
|
|
660 |
" 174 0.6565 0.0005 0.18s\n", |
|
|
661 |
" 175 0.6641 0.0002 0.18s\n", |
|
|
662 |
" 176 0.6537 0.0001 0.18s\n", |
|
|
663 |
" 177 0.6522 0.0007 0.18s\n", |
|
|
664 |
" 178 0.6557 0.0003 0.18s\n", |
|
|
665 |
" 179 0.6459 0.0001 0.18s\n", |
|
|
666 |
" 180 0.6608 0.0004 0.18s\n", |
|
|
667 |
" 181 0.6537 0.0003 0.17s\n", |
|
|
668 |
" 182 0.6645 0.0005 0.17s\n", |
|
|
669 |
" 183 0.6464 -0.0000 0.17s\n", |
|
|
670 |
" 184 0.6599 0.0004 0.17s\n", |
|
|
671 |
" 185 0.6458 0.0002 0.17s\n", |
|
|
672 |
" 186 0.6455 -0.0002 0.17s\n", |
|
|
673 |
" 187 0.6405 0.0000 0.17s\n", |
|
|
674 |
" 188 0.6539 0.0004 0.16s\n", |
|
|
675 |
" 189 0.6524 0.0004 0.16s\n", |
|
|
676 |
" 190 0.6436 0.0003 0.16s\n", |
|
|
677 |
" 191 0.6398 0.0002 0.16s\n", |
|
|
678 |
" 192 0.6367 0.0001 0.16s\n", |
|
|
679 |
" 193 0.6388 -0.0001 0.16s\n", |
|
|
680 |
" 194 0.6433 -0.0001 0.16s\n", |
|
|
681 |
" 195 0.6523 0.0005 0.15s\n", |
|
|
682 |
" 196 0.6532 0.0001 0.15s\n", |
|
|
683 |
" 197 0.6379 0.0002 0.15s\n", |
|
|
684 |
" 198 0.6368 0.0004 0.15s\n", |
|
|
685 |
" 199 0.6504 0.0003 0.15s\n", |
|
|
686 |
" 200 0.6426 0.0001 0.15s\n", |
|
|
687 |
" 201 0.6413 0.0005 0.15s\n", |
|
|
688 |
" 202 0.6502 0.0003 0.14s\n", |
|
|
689 |
" 203 0.6343 0.0000 0.14s\n", |
|
|
690 |
" 204 0.6324 0.0001 0.14s\n", |
|
|
691 |
" 205 0.6513 0.0005 0.14s\n", |
|
|
692 |
" 206 0.6427 0.0004 0.14s\n", |
|
|
693 |
" 207 0.6337 0.0003 0.14s\n", |
|
|
694 |
" 208 0.6403 0.0000 0.13s\n", |
|
|
695 |
" 209 0.6307 0.0001 0.13s\n", |
|
|
696 |
" 210 0.6276 -0.0001 0.13s\n", |
|
|
697 |
" 211 0.6454 0.0002 0.13s\n", |
|
|
698 |
" 212 0.6363 0.0004 0.13s\n", |
|
|
699 |
" 213 0.6509 0.0003 0.13s\n", |
|
|
700 |
" 214 0.6260 0.0004 0.13s\n", |
|
|
701 |
" 215 0.6161 0.0001 0.12s\n", |
|
|
702 |
" 216 0.6354 0.0003 0.12s\n", |
|
|
703 |
" 217 0.6405 0.0002 0.12s\n", |
|
|
704 |
" 218 0.6444 0.0002 0.12s\n", |
|
|
705 |
" 219 0.6325 0.0002 0.12s\n", |
|
|
706 |
" 220 0.6337 0.0001 0.12s\n", |
|
|
707 |
" 221 0.6348 0.0002 0.12s\n", |
|
|
708 |
" 222 0.6345 0.0003 0.11s\n", |
|
|
709 |
" 223 0.6390 0.0002 0.11s\n", |
|
|
710 |
" 224 0.6356 0.0003 0.11s\n", |
|
|
711 |
" 225 0.6464 0.0001 0.11s\n", |
|
|
712 |
" 226 0.6336 0.0004 0.11s\n", |
|
|
713 |
" 227 0.6352 -0.0000 0.11s\n", |
|
|
714 |
" 228 0.6185 0.0003 0.10s\n", |
|
|
715 |
" 229 0.6273 -0.0003 0.10s\n", |
|
|
716 |
" 230 0.6366 0.0005 0.10s\n", |
|
|
717 |
" 231 0.6147 0.0000 0.10s\n", |
|
|
718 |
" 232 0.6228 -0.0000 0.10s\n", |
|
|
719 |
" 233 0.6246 0.0003 0.10s\n", |
|
|
720 |
" 234 0.6331 0.0002 0.10s\n", |
|
|
721 |
" 235 0.6088 -0.0002 0.09s\n", |
|
|
722 |
" 236 0.6165 0.0003 0.09s\n", |
|
|
723 |
" 237 0.6198 0.0000 0.09s\n", |
|
|
724 |
" 238 0.6173 -0.0000 0.09s\n", |
|
|
725 |
" 239 0.6259 0.0002 0.09s\n", |
|
|
726 |
" 240 0.6080 -0.0001 0.09s\n", |
|
|
727 |
" 241 0.6144 0.0002 0.09s\n", |
|
|
728 |
" 242 0.6291 0.0001 0.08s\n", |
|
|
729 |
" 243 0.6195 -0.0000 0.08s\n", |
|
|
730 |
" 244 0.6171 0.0002 0.08s\n", |
|
|
731 |
" 245 0.6317 0.0002 0.08s\n", |
|
|
732 |
" 246 0.6216 0.0005 0.08s\n", |
|
|
733 |
" 247 0.6312 0.0004 0.08s\n", |
|
|
734 |
" 248 0.6284 0.0002 0.08s\n", |
|
|
735 |
" 249 0.6250 0.0003 0.07s\n", |
|
|
736 |
" 250 0.6111 -0.0001 0.07s\n", |
|
|
737 |
" 251 0.6137 -0.0002 0.07s\n", |
|
|
738 |
" 252 0.5821 -0.0006 0.07s\n", |
|
|
739 |
" 253 0.5978 0.0001 0.07s\n", |
|
|
740 |
" 254 0.6033 -0.0001 0.07s\n", |
|
|
741 |
" 255 0.6289 0.0001 0.07s\n", |
|
|
742 |
" 256 0.6091 0.0002 0.06s\n", |
|
|
743 |
" 257 0.6303 0.0000 0.06s\n", |
|
|
744 |
" 258 0.6193 0.0000 0.06s\n", |
|
|
745 |
" 259 0.6032 -0.0003 0.06s\n", |
|
|
746 |
" 260 0.6131 0.0001 0.06s\n", |
|
|
747 |
" 261 0.6177 0.0001 0.06s\n", |
|
|
748 |
" 262 0.6083 0.0002 0.06s\n", |
|
|
749 |
" 263 0.6053 -0.0000 0.05s\n", |
|
|
750 |
" 264 0.6086 0.0000 0.05s\n", |
|
|
751 |
" 265 0.6062 0.0000 0.05s\n", |
|
|
752 |
" 266 0.6049 -0.0002 0.05s\n", |
|
|
753 |
" 267 0.6063 -0.0001 0.05s\n", |
|
|
754 |
" 268 0.6054 0.0001 0.05s\n", |
|
|
755 |
" 269 0.5987 -0.0002 0.05s\n", |
|
|
756 |
" 270 0.6155 -0.0000 0.04s\n", |
|
|
757 |
" 271 0.6105 0.0003 0.04s\n", |
|
|
758 |
" 272 0.6154 0.0000 0.04s\n", |
|
|
759 |
" 273 0.6194 0.0002 0.04s\n", |
|
|
760 |
" 274 0.6078 -0.0002 0.04s\n", |
|
|
761 |
" 275 0.6019 0.0002 0.04s\n", |
|
|
762 |
" 276 0.6044 0.0003 0.03s\n", |
|
|
763 |
" 277 0.6129 0.0002 0.03s\n", |
|
|
764 |
" 278 0.6020 0.0000 0.03s\n", |
|
|
765 |
" 279 0.5923 0.0002 0.03s\n", |
|
|
766 |
" 280 0.6285 0.0000 0.03s\n", |
|
|
767 |
" 281 0.5960 -0.0001 0.03s\n", |
|
|
768 |
" 282 0.5949 -0.0001 0.03s\n", |
|
|
769 |
" 283 0.5965 -0.0006 0.02s\n", |
|
|
770 |
" 284 0.6116 -0.0002 0.02s\n", |
|
|
771 |
" 285 0.6052 0.0001 0.02s\n", |
|
|
772 |
" 286 0.6111 0.0003 0.02s\n", |
|
|
773 |
" 287 0.6103 0.0001 0.02s\n", |
|
|
774 |
" 288 0.6007 -0.0001 0.02s\n", |
|
|
775 |
" 289 0.5945 -0.0002 0.02s\n", |
|
|
776 |
" 290 0.6092 -0.0004 0.01s\n", |
|
|
777 |
" 291 0.6046 0.0002 0.01s\n", |
|
|
778 |
" 292 0.6077 -0.0001 0.01s\n", |
|
|
779 |
" 293 0.5816 -0.0003 0.01s\n", |
|
|
780 |
" 294 0.6066 0.0000 0.01s\n", |
|
|
781 |
" 295 0.5911 -0.0001 0.01s\n", |
|
|
782 |
" 296 0.5932 -0.0004 0.01s\n", |
|
|
783 |
" 297 0.5987 0.0003 0.00s\n", |
|
|
784 |
" 298 0.6014 0.0002 0.00s\n", |
|
|
785 |
" 299 0.6050 -0.0002 0.00s\n", |
|
|
786 |
" 300 0.5931 -0.0001 0.00s\n", |
|
|
787 |
" Iter Train Loss OOB Improve Remaining Time \n", |
|
|
788 |
" 1 0.9954 0.0040 0.45s\n", |
|
|
789 |
" 2 0.9911 0.0048 0.49s\n", |
|
|
790 |
" 3 0.9871 0.0043 0.47s\n", |
|
|
791 |
" 4 0.9820 0.0036 0.47s\n", |
|
|
792 |
" 5 0.9785 0.0042 0.46s\n", |
|
|
793 |
" 6 0.9741 0.0041 0.45s\n", |
|
|
794 |
" 7 0.9698 0.0041 0.45s\n", |
|
|
795 |
" 8 0.9658 0.0037 0.44s\n", |
|
|
796 |
" 9 0.9614 0.0038 0.44s\n", |
|
|
797 |
" 10 0.9581 0.0040 0.44s\n", |
|
|
798 |
" 11 0.9525 0.0035 0.44s\n", |
|
|
799 |
" 12 0.9493 0.0042 0.48s\n", |
|
|
800 |
" 13 0.9445 0.0031 0.49s\n", |
|
|
801 |
" 14 0.9398 0.0037 0.50s\n", |
|
|
802 |
" 15 0.9366 0.0033 0.49s\n", |
|
|
803 |
" 16 0.9317 0.0033 0.49s\n", |
|
|
804 |
" 17 0.9289 0.0031 0.48s\n", |
|
|
805 |
" 18 0.9247 0.0035 0.48s\n", |
|
|
806 |
" 19 0.9230 0.0039 0.47s\n", |
|
|
807 |
" 20 0.9172 0.0033 0.47s\n", |
|
|
808 |
" 21 0.9154 0.0036 0.46s\n", |
|
|
809 |
" 22 0.9094 0.0027 0.46s\n", |
|
|
810 |
" 23 0.9106 0.0033 0.46s\n", |
|
|
811 |
" 24 0.9068 0.0038 0.45s\n", |
|
|
812 |
" 25 0.9009 0.0033 0.45s\n", |
|
|
813 |
" 26 0.8949 0.0032 0.45s\n", |
|
|
814 |
" 27 0.8942 0.0033 0.44s\n", |
|
|
815 |
" 28 0.8904 0.0031 0.44s\n", |
|
|
816 |
" 29 0.8887 0.0029 0.44s\n", |
|
|
817 |
" 30 0.8846 0.0029 0.43s\n", |
|
|
818 |
" 31 0.8787 0.0026 0.43s\n", |
|
|
819 |
" 32 0.8790 0.0028 0.43s\n", |
|
|
820 |
" 33 0.8764 0.0031 0.43s\n", |
|
|
821 |
" 34 0.8711 0.0030 0.42s\n", |
|
|
822 |
" 35 0.8676 0.0025 0.42s\n", |
|
|
823 |
" 36 0.8630 0.0027 0.42s\n", |
|
|
824 |
" 37 0.8622 0.0029 0.42s\n", |
|
|
825 |
" 38 0.8594 0.0026 0.41s\n", |
|
|
826 |
" 39 0.8531 0.0024 0.41s\n", |
|
|
827 |
" 40 0.8564 0.0031 0.41s\n", |
|
|
828 |
" 41 0.8448 0.0023 0.41s\n", |
|
|
829 |
" 42 0.8540 0.0033 0.40s\n", |
|
|
830 |
" 43 0.8461 0.0026 0.40s\n", |
|
|
831 |
" 44 0.8394 0.0023 0.40s\n", |
|
|
832 |
" 45 0.8393 0.0023 0.40s\n", |
|
|
833 |
" 46 0.8346 0.0020 0.39s\n", |
|
|
834 |
" 47 0.8359 0.0024 0.39s\n", |
|
|
835 |
" 48 0.8343 0.0027 0.39s\n", |
|
|
836 |
" 49 0.8240 0.0020 0.39s\n", |
|
|
837 |
" 50 0.8289 0.0022 0.39s\n", |
|
|
838 |
" 51 0.8209 0.0020 0.38s\n", |
|
|
839 |
" 52 0.8221 0.0024 0.38s\n", |
|
|
840 |
" 53 0.8189 0.0022 0.38s\n", |
|
|
841 |
" 54 0.8248 0.0029 0.38s\n", |
|
|
842 |
" 55 0.8113 0.0017 0.38s\n", |
|
|
843 |
" 56 0.8113 0.0019 0.37s\n", |
|
|
844 |
" 57 0.8052 0.0016 0.37s\n", |
|
|
845 |
" 58 0.8026 0.0017 0.37s\n", |
|
|
846 |
" 59 0.8027 0.0018 0.37s\n", |
|
|
847 |
" 60 0.8011 0.0018 0.37s\n", |
|
|
848 |
" 61 0.7980 0.0023 0.37s\n", |
|
|
849 |
" 62 0.7938 0.0021 0.36s\n", |
|
|
850 |
" 63 0.7976 0.0017 0.36s\n", |
|
|
851 |
" 64 0.7965 0.0024 0.36s\n", |
|
|
852 |
" 65 0.7948 0.0020 0.36s\n", |
|
|
853 |
" 66 0.7966 0.0021 0.36s\n", |
|
|
854 |
" 67 0.7842 0.0019 0.36s\n", |
|
|
855 |
" 68 0.7801 0.0019 0.36s\n", |
|
|
856 |
" 69 0.7747 0.0017 0.35s\n", |
|
|
857 |
" 70 0.7743 0.0018 0.35s\n", |
|
|
858 |
" 71 0.7786 0.0019 0.35s\n", |
|
|
859 |
" 72 0.7731 0.0021 0.35s\n", |
|
|
860 |
" 73 0.7814 0.0019 0.35s\n", |
|
|
861 |
" 74 0.7820 0.0021 0.34s\n", |
|
|
862 |
" 75 0.7609 0.0014 0.34s\n", |
|
|
863 |
" 76 0.7611 0.0015 0.34s\n", |
|
|
864 |
" 77 0.7674 0.0011 0.34s\n", |
|
|
865 |
" 78 0.7658 0.0016 0.34s\n", |
|
|
866 |
" 79 0.7664 0.0016 0.34s\n", |
|
|
867 |
" 80 0.7640 0.0019 0.33s\n", |
|
|
868 |
" 81 0.7581 0.0017 0.33s\n", |
|
|
869 |
" 82 0.7513 0.0009 0.33s\n", |
|
|
870 |
" 83 0.7522 0.0013 0.33s\n", |
|
|
871 |
" 84 0.7586 0.0016 0.33s\n", |
|
|
872 |
" 85 0.7490 0.0014 0.32s\n", |
|
|
873 |
" 86 0.7401 0.0011 0.32s\n", |
|
|
874 |
" 87 0.7410 0.0010 0.32s\n", |
|
|
875 |
" 88 0.7445 0.0015 0.31s\n", |
|
|
876 |
" 89 0.7398 0.0008 0.31s\n", |
|
|
877 |
" 90 0.7415 0.0010 0.31s\n", |
|
|
878 |
" 91 0.7467 0.0017 0.31s\n", |
|
|
879 |
" 92 0.7424 0.0014 0.31s\n", |
|
|
880 |
" 93 0.7374 0.0012 0.30s\n", |
|
|
881 |
" 94 0.7474 0.0014 0.30s\n", |
|
|
882 |
" 95 0.7366 0.0010 0.30s\n", |
|
|
883 |
" 96 0.7397 0.0014 0.30s\n", |
|
|
884 |
" 97 0.7325 0.0010 0.30s\n", |
|
|
885 |
" 98 0.7311 0.0009 0.29s\n", |
|
|
886 |
" 99 0.7350 0.0015 0.29s\n", |
|
|
887 |
" 100 0.7218 0.0007 0.29s\n", |
|
|
888 |
" 101 0.7172 0.0008 0.29s\n", |
|
|
889 |
" 102 0.7259 0.0014 0.28s\n", |
|
|
890 |
" 103 0.7236 0.0012 0.28s\n", |
|
|
891 |
" 104 0.7165 0.0008 0.28s\n", |
|
|
892 |
" 105 0.7236 0.0011 0.28s\n", |
|
|
893 |
" 106 0.7274 0.0012 0.28s\n", |
|
|
894 |
" 107 0.7200 0.0011 0.28s\n", |
|
|
895 |
" 108 0.7232 0.0009 0.27s\n", |
|
|
896 |
" 109 0.7197 0.0008 0.27s\n", |
|
|
897 |
" 110 0.7191 0.0012 0.27s\n", |
|
|
898 |
" 111 0.7189 0.0008 0.27s\n", |
|
|
899 |
" 112 0.7148 0.0008 0.27s\n", |
|
|
900 |
" 113 0.7095 0.0006 0.27s\n", |
|
|
901 |
" 114 0.7094 0.0007 0.27s\n", |
|
|
902 |
" 115 0.7096 0.0007 0.27s\n", |
|
|
903 |
" 116 0.7105 0.0010 0.27s\n", |
|
|
904 |
" 117 0.7001 0.0002 0.27s\n", |
|
|
905 |
" 118 0.7137 0.0010 0.27s\n", |
|
|
906 |
" 119 0.7095 0.0008 0.26s\n", |
|
|
907 |
" 120 0.6956 0.0008 0.26s\n", |
|
|
908 |
" 121 0.7000 0.0007 0.26s\n", |
|
|
909 |
" 122 0.7034 0.0006 0.26s\n", |
|
|
910 |
" 123 0.7033 0.0007 0.26s\n", |
|
|
911 |
" 124 0.7032 0.0008 0.26s\n", |
|
|
912 |
" 125 0.6959 0.0008 0.26s\n", |
|
|
913 |
" 126 0.6985 0.0006 0.26s\n", |
|
|
914 |
" 127 0.6811 0.0006 0.26s\n", |
|
|
915 |
" 128 0.6959 0.0005 0.25s\n", |
|
|
916 |
" 129 0.6947 0.0004 0.25s\n", |
|
|
917 |
" 130 0.7033 0.0012 0.25s\n", |
|
|
918 |
" 131 0.7061 0.0008 0.25s\n", |
|
|
919 |
" 132 0.6855 0.0006 0.25s\n", |
|
|
920 |
" 133 0.6920 0.0006 0.25s\n", |
|
|
921 |
" 134 0.6890 0.0005 0.25s\n", |
|
|
922 |
" 135 0.6812 0.0008 0.24s\n", |
|
|
923 |
" 136 0.6939 0.0009 0.24s\n", |
|
|
924 |
" 137 0.6912 0.0012 0.24s\n", |
|
|
925 |
" 138 0.6889 0.0004 0.24s\n", |
|
|
926 |
" 139 0.6892 0.0007 0.24s\n", |
|
|
927 |
" 140 0.6845 0.0010 0.24s\n", |
|
|
928 |
" 141 0.6806 0.0006 0.23s\n", |
|
|
929 |
" 142 0.6986 0.0005 0.23s\n", |
|
|
930 |
" 143 0.6684 0.0005 0.23s\n", |
|
|
931 |
" 144 0.6826 0.0008 0.23s\n", |
|
|
932 |
" 145 0.6711 0.0001 0.23s\n", |
|
|
933 |
" 146 0.6820 0.0010 0.23s\n", |
|
|
934 |
" 147 0.6850 0.0007 0.23s\n", |
|
|
935 |
" 148 0.6721 0.0007 0.22s\n", |
|
|
936 |
" 149 0.6625 0.0003 0.22s\n", |
|
|
937 |
" 150 0.6853 0.0008 0.22s\n", |
|
|
938 |
" 151 0.6715 0.0005 0.22s\n", |
|
|
939 |
" 152 0.6783 0.0005 0.22s\n", |
|
|
940 |
" 153 0.6714 0.0005 0.22s\n", |
|
|
941 |
" 154 0.6705 0.0005 0.21s\n", |
|
|
942 |
" 155 0.6849 0.0006 0.21s\n", |
|
|
943 |
" 156 0.6588 0.0003 0.21s\n", |
|
|
944 |
" 157 0.6761 0.0005 0.21s\n", |
|
|
945 |
" 158 0.6696 0.0004 0.21s\n", |
|
|
946 |
" 159 0.6665 0.0008 0.21s\n", |
|
|
947 |
" 160 0.6735 0.0005 0.21s\n", |
|
|
948 |
" 161 0.6624 0.0008 0.20s\n", |
|
|
949 |
" 162 0.6774 0.0005 0.20s\n", |
|
|
950 |
" 163 0.6715 0.0005 0.20s\n", |
|
|
951 |
" 164 0.6672 0.0006 0.20s\n", |
|
|
952 |
" 165 0.6601 0.0006 0.20s\n", |
|
|
953 |
" 166 0.6642 0.0004 0.20s\n", |
|
|
954 |
" 167 0.6696 0.0003 0.20s\n", |
|
|
955 |
" 168 0.6487 -0.0000 0.19s\n", |
|
|
956 |
" 169 0.6523 0.0005 0.19s\n", |
|
|
957 |
" 170 0.6512 0.0001 0.19s\n", |
|
|
958 |
" 171 0.6529 0.0005 0.19s\n", |
|
|
959 |
" 172 0.6588 0.0004 0.19s\n", |
|
|
960 |
" 173 0.6520 0.0004 0.19s\n", |
|
|
961 |
" 174 0.6565 0.0005 0.18s\n", |
|
|
962 |
" 175 0.6641 0.0002 0.18s\n", |
|
|
963 |
" 176 0.6537 0.0001 0.18s\n", |
|
|
964 |
" 177 0.6522 0.0007 0.18s\n", |
|
|
965 |
" 178 0.6557 0.0003 0.18s\n", |
|
|
966 |
" 179 0.6459 0.0001 0.18s\n", |
|
|
967 |
" 180 0.6608 0.0004 0.18s\n", |
|
|
968 |
" 181 0.6537 0.0003 0.17s\n", |
|
|
969 |
" 182 0.6645 0.0005 0.17s\n", |
|
|
970 |
" 183 0.6464 -0.0000 0.17s\n", |
|
|
971 |
" 184 0.6599 0.0004 0.17s\n", |
|
|
972 |
" 185 0.6458 0.0002 0.17s\n", |
|
|
973 |
" 186 0.6455 -0.0002 0.17s\n", |
|
|
974 |
" 187 0.6405 0.0000 0.17s\n", |
|
|
975 |
" 188 0.6539 0.0004 0.16s\n", |
|
|
976 |
" 189 0.6524 0.0004 0.16s\n", |
|
|
977 |
" 190 0.6436 0.0003 0.16s\n", |
|
|
978 |
" 191 0.6398 0.0002 0.16s\n", |
|
|
979 |
" 192 0.6367 0.0001 0.16s\n", |
|
|
980 |
" 193 0.6388 -0.0001 0.16s\n", |
|
|
981 |
" 194 0.6433 -0.0001 0.15s\n", |
|
|
982 |
" 195 0.6523 0.0005 0.15s\n", |
|
|
983 |
" 196 0.6532 0.0001 0.15s\n", |
|
|
984 |
" 197 0.6379 0.0002 0.15s\n", |
|
|
985 |
" 198 0.6368 0.0004 0.15s\n", |
|
|
986 |
" 199 0.6504 0.0003 0.15s\n", |
|
|
987 |
" 200 0.6426 0.0001 0.15s\n", |
|
|
988 |
" 201 0.6413 0.0005 0.14s\n", |
|
|
989 |
" 202 0.6502 0.0003 0.14s\n", |
|
|
990 |
" 203 0.6343 0.0000 0.14s\n", |
|
|
991 |
" 204 0.6324 0.0001 0.14s\n", |
|
|
992 |
" 205 0.6513 0.0005 0.14s\n", |
|
|
993 |
" 206 0.6427 0.0004 0.14s\n", |
|
|
994 |
" 207 0.6337 0.0003 0.14s\n", |
|
|
995 |
" 208 0.6403 0.0000 0.14s\n", |
|
|
996 |
" 209 0.6307 0.0001 0.14s\n", |
|
|
997 |
" 210 0.6276 -0.0001 0.13s\n", |
|
|
998 |
" 211 0.6454 0.0002 0.13s\n", |
|
|
999 |
" 212 0.6363 0.0004 0.13s\n", |
|
|
1000 |
" 213 0.6509 0.0003 0.13s\n", |
|
|
1001 |
" 214 0.6260 0.0004 0.13s\n", |
|
|
1002 |
" 215 0.6161 0.0001 0.13s\n", |
|
|
1003 |
" 216 0.6354 0.0003 0.13s\n", |
|
|
1004 |
" 217 0.6405 0.0002 0.12s\n", |
|
|
1005 |
" 218 0.6444 0.0002 0.12s\n", |
|
|
1006 |
" 219 0.6325 0.0002 0.12s\n", |
|
|
1007 |
" 220 0.6337 0.0001 0.12s\n", |
|
|
1008 |
" 221 0.6348 0.0002 0.12s\n", |
|
|
1009 |
" 222 0.6345 0.0003 0.12s\n", |
|
|
1010 |
" 223 0.6390 0.0002 0.11s\n", |
|
|
1011 |
" 224 0.6356 0.0003 0.11s\n", |
|
|
1012 |
" 225 0.6464 0.0001 0.11s\n", |
|
|
1013 |
" 226 0.6336 0.0004 0.11s\n", |
|
|
1014 |
" 227 0.6352 -0.0000 0.11s\n", |
|
|
1015 |
" 228 0.6185 0.0003 0.11s\n", |
|
|
1016 |
" 229 0.6273 -0.0003 0.11s\n", |
|
|
1017 |
" 230 0.6366 0.0005 0.11s\n", |
|
|
1018 |
" 231 0.6147 0.0000 0.10s\n", |
|
|
1019 |
" 232 0.6228 -0.0000 0.10s\n", |
|
|
1020 |
" 233 0.6246 0.0003 0.10s\n", |
|
|
1021 |
" 234 0.6331 0.0002 0.10s\n", |
|
|
1022 |
" 235 0.6088 -0.0002 0.10s\n", |
|
|
1023 |
" 236 0.6165 0.0003 0.10s\n", |
|
|
1024 |
" 237 0.6198 0.0000 0.09s\n", |
|
|
1025 |
" 238 0.6173 -0.0000 0.09s\n", |
|
|
1026 |
" 239 0.6259 0.0002 0.09s\n", |
|
|
1027 |
" 240 0.6080 -0.0001 0.09s\n", |
|
|
1028 |
" 241 0.6144 0.0002 0.09s\n", |
|
|
1029 |
" 242 0.6291 0.0001 0.09s\n", |
|
|
1030 |
" 243 0.6195 -0.0000 0.09s\n", |
|
|
1031 |
" 244 0.6171 0.0002 0.08s\n", |
|
|
1032 |
" 245 0.6317 0.0002 0.08s\n", |
|
|
1033 |
" 246 0.6216 0.0005 0.08s\n", |
|
|
1034 |
" 247 0.6312 0.0004 0.08s\n", |
|
|
1035 |
" 248 0.6284 0.0002 0.08s\n", |
|
|
1036 |
" 249 0.6250 0.0003 0.08s\n", |
|
|
1037 |
" 250 0.6111 -0.0001 0.08s\n", |
|
|
1038 |
" 251 0.6137 -0.0002 0.07s\n", |
|
|
1039 |
" 252 0.5821 -0.0006 0.07s\n", |
|
|
1040 |
" 253 0.5978 0.0001 0.07s\n", |
|
|
1041 |
" 254 0.6033 -0.0001 0.07s\n", |
|
|
1042 |
" 255 0.6289 0.0001 0.07s\n", |
|
|
1043 |
" 256 0.6091 0.0002 0.07s\n", |
|
|
1044 |
" 257 0.6303 0.0000 0.07s\n", |
|
|
1045 |
" 258 0.6193 0.0000 0.06s\n", |
|
|
1046 |
" 259 0.6032 -0.0003 0.06s\n", |
|
|
1047 |
" 260 0.6131 0.0001 0.06s\n", |
|
|
1048 |
" 261 0.6177 0.0001 0.06s\n", |
|
|
1049 |
" 262 0.6083 0.0002 0.06s\n", |
|
|
1050 |
" 263 0.6053 -0.0000 0.06s\n", |
|
|
1051 |
" 264 0.6086 0.0000 0.05s\n", |
|
|
1052 |
" 265 0.6062 0.0000 0.05s\n", |
|
|
1053 |
" 266 0.6049 -0.0002 0.05s\n", |
|
|
1054 |
" 267 0.6063 -0.0001 0.05s\n", |
|
|
1055 |
" 268 0.6054 0.0001 0.05s\n", |
|
|
1056 |
" 269 0.5987 -0.0002 0.05s\n", |
|
|
1057 |
" 270 0.6155 -0.0000 0.04s\n", |
|
|
1058 |
" 271 0.6105 0.0003 0.04s\n", |
|
|
1059 |
" 272 0.6154 0.0000 0.04s\n", |
|
|
1060 |
" 273 0.6194 0.0002 0.04s\n", |
|
|
1061 |
" 274 0.6078 -0.0002 0.04s\n", |
|
|
1062 |
" 275 0.6019 0.0002 0.04s\n", |
|
|
1063 |
" 276 0.6044 0.0003 0.04s\n", |
|
|
1064 |
" 277 0.6129 0.0002 0.03s\n", |
|
|
1065 |
" 278 0.6020 0.0000 0.03s\n", |
|
|
1066 |
" 279 0.5923 0.0002 0.03s\n", |
|
|
1067 |
" 280 0.6285 0.0000 0.03s\n", |
|
|
1068 |
" 281 0.5960 -0.0001 0.03s\n", |
|
|
1069 |
" 282 0.5949 -0.0001 0.03s\n", |
|
|
1070 |
" 283 0.5965 -0.0006 0.03s\n", |
|
|
1071 |
" 284 0.6116 -0.0002 0.02s\n", |
|
|
1072 |
" 285 0.6052 0.0001 0.02s\n", |
|
|
1073 |
" 286 0.6111 0.0003 0.02s\n", |
|
|
1074 |
" 287 0.6103 0.0001 0.02s\n", |
|
|
1075 |
" 288 0.6007 -0.0001 0.02s\n", |
|
|
1076 |
" 289 0.5945 -0.0002 0.02s\n", |
|
|
1077 |
" 290 0.6092 -0.0004 0.01s\n", |
|
|
1078 |
" 291 0.6046 0.0002 0.01s\n", |
|
|
1079 |
" 292 0.6077 -0.0001 0.01s\n", |
|
|
1080 |
" 293 0.5816 -0.0003 0.01s\n", |
|
|
1081 |
" 294 0.6066 0.0000 0.01s\n", |
|
|
1082 |
" 295 0.5911 -0.0001 0.01s\n", |
|
|
1083 |
" 296 0.5932 -0.0004 0.01s\n", |
|
|
1084 |
" 297 0.5987 0.0003 0.00s\n", |
|
|
1085 |
" 298 0.6014 0.0002 0.00s\n", |
|
|
1086 |
" 299 0.6050 -0.0002 0.00s\n", |
|
|
1087 |
" 300 0.5931 -0.0001 0.00s\n", |
|
|
1088 |
"Best Parameters for NSCLC: {'gradient_boosting__warm_start': True, 'gradient_boosting__verbose': 2, 'gradient_boosting__validation_fraction': 0.2, 'gradient_boosting__tol': 0.0001, 'gradient_boosting__subsample': 0.8, 'gradient_boosting__random_state': 100, 'gradient_boosting__n_iter_no_change': 20, 'gradient_boosting__n_estimators': 300, 'gradient_boosting__min_weight_fraction_leaf': 0.0, 'gradient_boosting__min_samples_split': 4, 'gradient_boosting__min_samples_leaf': 1, 'gradient_boosting__min_impurity_decrease': 0.0, 'gradient_boosting__max_leaf_nodes': None, 'gradient_boosting__max_features': 'log2', 'gradient_boosting__max_depth': 3, 'gradient_boosting__loss': 'exponential', 'gradient_boosting__learning_rate': 0.01, 'gradient_boosting__init': 'zero', 'gradient_boosting__criterion': 'squared_error', 'gradient_boosting__ccp_alpha': 0.0}\n", |
|
|
1089 |
"Best F1 Score for NSCLC: 0.8471398246548704\n", |
|
|
1090 |
"Classification Report for NSCLC (Validation Data):\n", |
|
|
1091 |
" precision recall f1-score support\n", |
|
|
1092 |
"\n", |
|
|
1093 |
" 0.0 0.82 0.82 0.82 142\n", |
|
|
1094 |
" 1.0 0.81 0.81 0.81 135\n", |
|
|
1095 |
"\n", |
|
|
1096 |
" accuracy 0.82 277\n", |
|
|
1097 |
" macro avg 0.82 0.82 0.82 277\n", |
|
|
1098 |
"weighted avg 0.82 0.82 0.82 277\n", |
|
|
1099 |
"\n" |
|
|
1100 |
] |
|
|
1101 |
} |
|
|
1102 |
], |
|
|
1103 |
"source": [ |
|
|
1104 |
"#evaluate for NSCLC\n", |
|
|
1105 |
"best_params_nsclc, best_score_nsclc, report_nsclc = evaluate_gradient_boosting(xtrain_nsclc, ytrain_nsclc, xtest_nsclc, ytest_nsclc, param_grid_gradient_boosting)\n", |
|
|
1106 |
"print(\"Best Parameters for NSCLC:\", best_params_nsclc)\n", |
|
|
1107 |
"print(\"Best F1 Score for NSCLC:\", best_score_nsclc)\n", |
|
|
1108 |
"print(\"Classification Report for NSCLC (Validation Data):\\n\", report_nsclc)" |
|
|
1109 |
], |
|
|
1110 |
"id": "28d9744d-ee67-4ba5-aa2b-d1ec35c0807f" |
|
|
1111 |
}, |
|
|
1112 |
{ |
|
|
1113 |
"cell_type": "code", |
|
|
1114 |
"execution_count": 98, |
|
|
1115 |
"metadata": { |
|
|
1116 |
"id": "9ef05318-c378-4406-a597-f2d5a107324d", |
|
|
1117 |
"colab": { |
|
|
1118 |
"base_uri": "https://localhost:8080/" |
|
|
1119 |
}, |
|
|
1120 |
"outputId": "51ef3bad-4e5e-4c2e-da27-f0bdd8537d07" |
|
|
1121 |
}, |
|
|
1122 |
"outputs": [ |
|
|
1123 |
{ |
|
|
1124 |
"output_type": "stream", |
|
|
1125 |
"name": "stdout", |
|
|
1126 |
"text": [ |
|
|
1127 |
"Fitting 10 folds for each of 1000 candidates, totalling 10000 fits\n", |
|
|
1128 |
" Iter Train Loss OOB Improve Remaining Time \n", |
|
|
1129 |
" 1 0.9527 0.0450 0.52s\n", |
|
|
1130 |
" 2 0.9162 0.0400 0.62s\n", |
|
|
1131 |
" 3 0.8864 0.0323 0.84s\n", |
|
|
1132 |
" 4 0.8462 0.0334 0.79s\n", |
|
|
1133 |
" 5 0.8214 0.0237 0.76s\n", |
|
|
1134 |
" 6 0.7984 0.0239 0.73s\n", |
|
|
1135 |
" 7 0.7773 0.0183 0.71s\n", |
|
|
1136 |
" 8 0.7377 0.0191 0.70s\n", |
|
|
1137 |
" 9 0.7321 0.0141 0.69s\n", |
|
|
1138 |
" 10 0.7272 0.0129 0.68s\n", |
|
|
1139 |
" 20 0.6332 0.0009 0.56s\n", |
|
|
1140 |
" 30 0.5971 0.0035 0.52s\n", |
|
|
1141 |
" 40 0.5579 -0.0011 0.52s\n", |
|
|
1142 |
" Iter Train Loss OOB Improve Remaining Time \n", |
|
|
1143 |
" 301 2300.6212 220.6920 0.14s\n", |
|
|
1144 |
" 302 2084.1484 164.0025 0.17s\n", |
|
|
1145 |
" 303 1671.1370 93.5445 0.18s\n", |
|
|
1146 |
" 304 1699.2235 284.4636 0.18s\n", |
|
|
1147 |
" 305 2127.2450 -39.8442 0.18s\n", |
|
|
1148 |
" 306 1125.9058 80.8256 0.17s\n", |
|
|
1149 |
" 307 1029.6187 294.3713 0.17s\n", |
|
|
1150 |
" 308 1863.1056 -76.3355 0.17s\n", |
|
|
1151 |
" 309 2092.8167 31.4775 0.17s\n", |
|
|
1152 |
" 310 1060.5321 -181.3740 0.17s\n", |
|
|
1153 |
" 320 903.3716 41.6736 0.13s\n", |
|
|
1154 |
"Best Parameters for SCLC: {'gradient_boosting__warm_start': True, 'gradient_boosting__verbose': 1, 'gradient_boosting__validation_fraction': 0.1, 'gradient_boosting__tol': 0.0001, 'gradient_boosting__subsample': 0.6, 'gradient_boosting__random_state': 100, 'gradient_boosting__n_iter_no_change': 10, 'gradient_boosting__n_estimators': 400, 'gradient_boosting__min_weight_fraction_leaf': 0.0, 'gradient_boosting__min_samples_split': 4, 'gradient_boosting__min_samples_leaf': 1, 'gradient_boosting__min_impurity_decrease': 0.0, 'gradient_boosting__max_leaf_nodes': None, 'gradient_boosting__max_features': 'sqrt', 'gradient_boosting__max_depth': 3, 'gradient_boosting__loss': 'exponential', 'gradient_boosting__learning_rate': 0.1, 'gradient_boosting__init': None, 'gradient_boosting__criterion': 'friedman_mse', 'gradient_boosting__ccp_alpha': 0.0}\n", |
|
|
1155 |
"Best F1 Score for SCLC: 0.8471492954496602\n", |
|
|
1156 |
"Classification Report for SCLC (Validation Data):\n", |
|
|
1157 |
" precision recall f1-score support\n", |
|
|
1158 |
"\n", |
|
|
1159 |
" 0.0 0.82 0.85 0.83 142\n", |
|
|
1160 |
" 1.0 0.84 0.80 0.82 135\n", |
|
|
1161 |
"\n", |
|
|
1162 |
" accuracy 0.83 277\n", |
|
|
1163 |
" macro avg 0.83 0.83 0.83 277\n", |
|
|
1164 |
"weighted avg 0.83 0.83 0.83 277\n", |
|
|
1165 |
"\n" |
|
|
1166 |
] |
|
|
1167 |
} |
|
|
1168 |
], |
|
|
1169 |
"source": [ |
|
|
1170 |
"#evaluate for SCLC\n", |
|
|
1171 |
"best_params_sclc, best_score_sclc, report_sclc = evaluate_gradient_boosting(xtrain_nsclc, ytrain_nsclc, xtest_nsclc, ytest_nsclc, param_grid_gradient_boosting)\n", |
|
|
1172 |
"print(\"Best Parameters for SCLC:\", best_params_sclc)\n", |
|
|
1173 |
"print(\"Best F1 Score for SCLC:\", best_score_sclc)\n", |
|
|
1174 |
"print(\"Classification Report for SCLC (Validation Data):\\n\", report_sclc)" |
|
|
1175 |
], |
|
|
1176 |
"id": "9ef05318-c378-4406-a597-f2d5a107324d" |
|
|
1177 |
} |
|
|
1178 |
], |
|
|
1179 |
"metadata": { |
|
|
1180 |
"colab": { |
|
|
1181 |
"provenance": [] |
|
|
1182 |
}, |
|
|
1183 |
"kernelspec": { |
|
|
1184 |
"display_name": "Python 3 (ipykernel)", |
|
|
1185 |
"language": "python", |
|
|
1186 |
"name": "python3" |
|
|
1187 |
}, |
|
|
1188 |
"language_info": { |
|
|
1189 |
"codemirror_mode": { |
|
|
1190 |
"name": "ipython", |
|
|
1191 |
"version": 3 |
|
|
1192 |
}, |
|
|
1193 |
"file_extension": ".py", |
|
|
1194 |
"mimetype": "text/x-python", |
|
|
1195 |
"name": "python", |
|
|
1196 |
"nbconvert_exporter": "python", |
|
|
1197 |
"pygments_lexer": "ipython3", |
|
|
1198 |
"version": "3.9.18" |
|
|
1199 |
} |
|
|
1200 |
}, |
|
|
1201 |
"nbformat": 4, |
|
|
1202 |
"nbformat_minor": 5 |
|
|
1203 |
} |