Show More
@@ -1,12 +1,16 b'' | |||
|
1 | 1 | { |
|
2 |
"metadata": { |
|
|
2 | "metadata": { | |
|
3 | "name": "tutorial" | |
|
4 | }, | |
|
3 | 5 | "nbformat": 3, |
|
6 | "nbformat_minor": 0, | |
|
4 | 7 | "worksheets": [ |
|
5 | 8 | { |
|
6 | 9 | "cells": [ |
|
7 | 10 | { |
|
8 | 11 | "cell_type": "heading", |
|
9 | 12 | "level": 1, |
|
13 | "metadata": {}, | |
|
10 | 14 | "source": [ |
|
11 | 15 | "An Introduction to machine learning with scikit-learn" |
|
12 | 16 | ] |
@@ -14,121 +18,134 b'' | |||
|
14 | 18 | { |
|
15 | 19 | "cell_type": "heading", |
|
16 | 20 | "level": 1, |
|
21 | "metadata": {}, | |
|
17 | 22 | "source": [ |
|
18 | 23 | "Section contents" |
|
19 | 24 | ] |
|
20 | 25 | }, |
|
21 | 26 | { |
|
22 | 27 | "cell_type": "markdown", |
|
28 | "metadata": {}, | |
|
23 | 29 | "source": [ |
|
24 | "In this section, we introduce the machine learning", | |
|
25 | "vocabulary that we use through-out scikit-learn and give a", | |
|
30 | "In this section, we introduce the machine learning\n", | |
|
31 | "vocabulary that we use through-out scikit-learn and give a\n", | |
|
26 | 32 | "simple learning example." |
|
27 | 33 | ] |
|
28 | 34 | }, |
|
29 | 35 | { |
|
30 | 36 | "cell_type": "heading", |
|
31 | 37 | "level": 2, |
|
38 | "metadata": {}, | |
|
32 | 39 | "source": [ |
|
33 | 40 | "Machine learning: the problem setting" |
|
34 | 41 | ] |
|
35 | 42 | }, |
|
36 | 43 | { |
|
37 | 44 | "cell_type": "markdown", |
|
38 | "source": [ | |
|
39 | "In general, a learning problem considers a set of n", | |
|
40 | "samples of", | |
|
41 | "data and try to predict properties of unknown data. If each sample is", | |
|
42 | "more than a single number, and for instance a multi-dimensional entry", | |
|
43 | "(aka multivariate", | |
|
44 | "data), is it said to have several attributes,", | |
|
45 | "metadata": {}, | |
|
46 | "source": [ | |
|
47 | "In general, a learning problem considers a set of n\n", | |
|
48 | "samples of\n", | |
|
49 | "data and try to predict properties of unknown data. If each sample is\n", | |
|
50 | "more than a single number, and for instance a multi-dimensional entry\n", | |
|
51 | "(aka multivariate\n", | |
|
52 | "data), is it said to have several attributes,\n", | |
|
45 | 53 | "or features." |
|
46 | 54 | ] |
|
47 | 55 | }, |
|
48 | 56 | { |
|
49 | 57 | "cell_type": "markdown", |
|
58 | "metadata": {}, | |
|
50 | 59 | "source": [ |
|
51 | 60 | "We can separate learning problems in a few large categories:" |
|
52 | 61 | ] |
|
53 | 62 | }, |
|
54 | 63 | { |
|
55 | 64 | "cell_type": "markdown", |
|
65 | "metadata": {}, | |
|
56 | 66 | "source": [ |
|
57 | "supervised learning,", | |
|
58 | "in which the data comes with additional attributes that we want to predict", | |
|
59 | "(:ref:`Click here <supervised-learning>`", | |
|
60 | "to go to the Scikit-Learn supervised learning page).This problem", | |
|
67 | "supervised learning,\n", | |
|
68 | "in which the data comes with additional attributes that we want to predict\n", | |
|
69 | "(:ref:`Click here <supervised-learning>`\n", | |
|
70 | "to go to the Scikit-Learn supervised learning page).This problem\n", | |
|
61 | 71 | "can be either:" |
|
62 | 72 | ] |
|
63 | 73 | }, |
|
64 | 74 | { |
|
65 | 75 | "cell_type": "markdown", |
|
66 | "source": [ | |
|
67 | "classification:", | |
|
68 | "samples belong to two or more classes and we", | |
|
69 | "want to learn from already labeled data how to predict the class", | |
|
70 | "of unlabeled data. An example of classification problem would", | |
|
71 | "be the digit recognition example, in which the aim is to assign", | |
|
72 | "each input vector to one of a finite number of discrete", | |
|
76 | "metadata": {}, | |
|
77 | "source": [ | |
|
78 | "classification:\n", | |
|
79 | "samples belong to two or more classes and we\n", | |
|
80 | "want to learn from already labeled data how to predict the class\n", | |
|
81 | "of unlabeled data. An example of classification problem would\n", | |
|
82 | "be the digit recognition example, in which the aim is to assign\n", | |
|
83 | "each input vector to one of a finite number of discrete\n", | |
|
73 | 84 | "categories." |
|
74 | 85 | ] |
|
75 | 86 | }, |
|
76 | 87 | { |
|
77 | 88 | "cell_type": "markdown", |
|
89 | "metadata": {}, | |
|
78 | 90 | "source": [ |
|
79 | "regression:", | |
|
80 | "if the desired output consists of one or more", | |
|
81 | "continuous variables, then the task is called regression. An", | |
|
82 | "example of a regression problem would be the prediction of the", | |
|
91 | "regression:\n", | |
|
92 | "if the desired output consists of one or more\n", | |
|
93 | "continuous variables, then the task is called regression. An\n", | |
|
94 | "example of a regression problem would be the prediction of the\n", | |
|
83 | 95 | "length of a salmon as a function of its age and weight." |
|
84 | 96 | ] |
|
85 | 97 | }, |
|
86 | 98 | { |
|
87 | 99 | "cell_type": "markdown", |
|
88 | "source": [ | |
|
89 | "unsupervised learning,", | |
|
90 | "in which the training data consists of a set of input vectors x", | |
|
91 | "without any corresponding target values. The goal in such problems", | |
|
92 | "may be to discover groups of similar examples within the data, where", | |
|
93 | "it is called clustering,", | |
|
94 | "or to determine the distribution of data within the input space, known as", | |
|
95 | "density estimation, or", | |
|
96 | "to project the data from a high-dimensional space down to two or thee", | |
|
97 | "dimensions for the purpose of visualization", | |
|
98 | "(:ref:`Click here <unsupervised-learning>`", | |
|
100 | "metadata": {}, | |
|
101 | "source": [ | |
|
102 | "unsupervised learning,\n", | |
|
103 | "in which the training data consists of a set of input vectors x\n", | |
|
104 | "without any corresponding target values. The goal in such problems\n", | |
|
105 | "may be to discover groups of similar examples within the data, where\n", | |
|
106 | "it is called clustering,\n", | |
|
107 | "or to determine the distribution of data within the input space, known as\n", | |
|
108 | "density estimation, or\n", | |
|
109 | "to project the data from a high-dimensional space down to two or thee\n", | |
|
110 | "dimensions for the purpose of visualization\n", | |
|
111 | "(:ref:`Click here <unsupervised-learning>`\n", | |
|
99 | 112 | "to go to the Scikit-Learn unsupervised learning page)." |
|
100 | 113 | ] |
|
101 | 114 | }, |
|
102 | 115 | { |
|
103 | 116 | "cell_type": "heading", |
|
104 | 117 | "level": 2, |
|
118 | "metadata": {}, | |
|
105 | 119 | "source": [ |
|
106 | 120 | "Training set and testing set" |
|
107 | 121 | ] |
|
108 | 122 | }, |
|
109 | 123 | { |
|
110 | 124 | "cell_type": "markdown", |
|
125 | "metadata": {}, | |
|
111 | 126 | "source": [ |
|
112 | "Machine learning is about learning some properties of a data set", | |
|
113 | "and applying them to new data. This is why a common practice in", | |
|
114 | "machine learning to evaluate an algorithm is to split the data", | |
|
115 | "at hand in two sets, one that we call a training set on which", | |
|
116 | "we learn data properties, and one that we call a testing set,", | |
|
127 | "Machine learning is about learning some properties of a data set\n", | |
|
128 | "and applying them to new data. This is why a common practice in\n", | |
|
129 | "machine learning to evaluate an algorithm is to split the data\n", | |
|
130 | "at hand in two sets, one that we call a training set on which\n", | |
|
131 | "we learn data properties, and one that we call a testing set,\n", | |
|
117 | 132 | "on which we test these properties." |
|
118 | 133 | ] |
|
119 | 134 | }, |
|
120 | 135 | { |
|
121 | 136 | "cell_type": "heading", |
|
122 | 137 | "level": 2, |
|
138 | "metadata": {}, | |
|
123 | 139 | "source": [ |
|
124 | 140 | "Loading an example dataset" |
|
125 | 141 | ] |
|
126 | 142 | }, |
|
127 | 143 | { |
|
128 | 144 | "cell_type": "markdown", |
|
145 | "metadata": {}, | |
|
129 | 146 | "source": [ |
|
130 | "scikit-learn comes with a few standard datasets, for instance the", | |
|
131 | "iris and digits", | |
|
147 | "scikit-learn comes with a few standard datasets, for instance the\n", | |
|
148 | "iris and digits\n", | |
|
132 | 149 | "datasets for classification and the boston house prices dataset for regression.:" |
|
133 | 150 | ] |
|
134 | 151 | }, |
@@ -136,28 +153,31 b'' | |||
|
136 | 153 | "cell_type": "code", |
|
137 | 154 | "collapsed": false, |
|
138 | 155 | "input": [ |
|
139 | "from sklearn import datasets", | |
|
140 | "iris = datasets.load_iris()", | |
|
156 | "from sklearn import datasets\n", | |
|
157 | "iris = datasets.load_iris()\n", | |
|
141 | 158 | "digits = datasets.load_digits()" |
|
142 | 159 | ], |
|
143 | 160 | "language": "python", |
|
161 | "metadata": {}, | |
|
144 | 162 | "outputs": [] |
|
145 | 163 | }, |
|
146 | 164 | { |
|
147 | 165 | "cell_type": "markdown", |
|
166 | "metadata": {}, | |
|
148 | 167 | "source": [ |
|
149 | "A dataset is a dictionary-like object that holds all the data and some", | |
|
150 | "metadata about the data. This data is stored in the .data member,", | |
|
151 | "which is a n_samples, n_features array. In the case of supervised", | |
|
152 | "problem, explanatory variables are stored in the .target member. More", | |
|
153 | "details on the different datasets can be found in the :ref:`dedicated", | |
|
168 | "A dataset is a dictionary-like object that holds all the data and some\n", | |
|
169 | "metadata about the data. This data is stored in the .data member,\n", | |
|
170 | "which is a n_samples, n_features array. In the case of supervised\n", | |
|
171 | "problem, explanatory variables are stored in the .target member. More\n", | |
|
172 | "details on the different datasets can be found in the :ref:`dedicated\n", | |
|
154 | 173 | "section <datasets>`." |
|
155 | 174 | ] |
|
156 | 175 | }, |
|
157 | 176 | { |
|
158 | 177 | "cell_type": "markdown", |
|
178 | "metadata": {}, | |
|
159 | 179 | "source": [ |
|
160 | "For instance, in the case of the digits dataset, digits.data gives", | |
|
180 | "For instance, in the case of the digits dataset, digits.data gives\n", | |
|
161 | 181 | "access to the features that can be used to classify the digits samples:" |
|
162 | 182 | ] |
|
163 | 183 | }, |
@@ -168,13 +188,15 b'' | |||
|
168 | 188 | "print digits.data # doctest: +NORMALIZE_WHITESPACE" |
|
169 | 189 | ], |
|
170 | 190 | "language": "python", |
|
191 | "metadata": {}, | |
|
171 | 192 | "outputs": [] |
|
172 | 193 | }, |
|
173 | 194 | { |
|
174 | 195 | "cell_type": "markdown", |
|
196 | "metadata": {}, | |
|
175 | 197 | "source": [ |
|
176 | "and digits.target gives the ground truth for the digit dataset, that", | |
|
177 | "is the number corresponding to each digit image that we are trying to", | |
|
198 | "and digits.target gives the ground truth for the digit dataset, that\n", | |
|
199 | "is the number corresponding to each digit image that we are trying to\n", | |
|
178 | 200 | "learn:" |
|
179 | 201 | ] |
|
180 | 202 | }, |
@@ -185,21 +207,24 b'' | |||
|
185 | 207 | "digits.target" |
|
186 | 208 | ], |
|
187 | 209 | "language": "python", |
|
210 | "metadata": {}, | |
|
188 | 211 | "outputs": [] |
|
189 | 212 | }, |
|
190 | 213 | { |
|
191 | 214 | "cell_type": "heading", |
|
192 | 215 | "level": 2, |
|
216 | "metadata": {}, | |
|
193 | 217 | "source": [ |
|
194 | 218 | "Shape of the data arrays" |
|
195 | 219 | ] |
|
196 | 220 | }, |
|
197 | 221 | { |
|
198 | 222 | "cell_type": "markdown", |
|
223 | "metadata": {}, | |
|
199 | 224 | "source": [ |
|
200 | "The data is always a 2D array, n_samples, n_features, although", | |
|
201 | "the original data may have had a different shape. In the case of the", | |
|
202 | "digits, each original sample is an image of shape 8, 8 and can be", | |
|
225 | "The data is always a 2D array, n_samples, n_features, although\n", | |
|
226 | "the original data may have had a different shape. In the case of the\n", | |
|
227 | "digits, each original sample is an image of shape 8, 8 and can be\n", | |
|
203 | 228 | "accessed using:" |
|
204 | 229 | ] |
|
205 | 230 | }, |
@@ -210,48 +235,54 b'' | |||
|
210 | 235 | "digits.images[0]" |
|
211 | 236 | ], |
|
212 | 237 | "language": "python", |
|
238 | "metadata": {}, | |
|
213 | 239 | "outputs": [] |
|
214 | 240 | }, |
|
215 | 241 | { |
|
216 | 242 | "cell_type": "markdown", |
|
243 | "metadata": {}, | |
|
217 | 244 | "source": [ |
|
218 | "The :ref:`simple example on this dataset", | |
|
219 | "<example_plot_digits_classification.py>` illustrates how starting", | |
|
220 | "from the original problem one can shape the data for consumption in", | |
|
245 | "The :ref:`simple example on this dataset\n", | |
|
246 | "<example_plot_digits_classification.py>` illustrates how starting\n", | |
|
247 | "from the original problem one can shape the data for consumption in\n", | |
|
221 | 248 | "the scikit-learn." |
|
222 | 249 | ] |
|
223 | 250 | }, |
|
224 | 251 | { |
|
225 | 252 | "cell_type": "heading", |
|
226 | 253 | "level": 2, |
|
254 | "metadata": {}, | |
|
227 | 255 | "source": [ |
|
228 | 256 | "Learning and Predicting" |
|
229 | 257 | ] |
|
230 | 258 | }, |
|
231 | 259 | { |
|
232 | 260 | "cell_type": "markdown", |
|
261 | "metadata": {}, | |
|
233 | 262 | "source": [ |
|
234 | "In the case of the digits dataset, the task is to predict the value of a", | |
|
235 | "hand-written digit from an image. We are given samples of each of the 10", | |
|
236 | "possible classes on which we fit an", | |
|
237 | "estimator to be able to predict", | |
|
263 | "In the case of the digits dataset, the task is to predict the value of a\n", | |
|
264 | "hand-written digit from an image. We are given samples of each of the 10\n", | |
|
265 | "possible classes on which we fit an\n", | |
|
266 | "estimator to be able to predict\n", | |
|
238 | 267 | "the labels corresponding to new data." |
|
239 | 268 | ] |
|
240 | 269 | }, |
|
241 | 270 | { |
|
242 | 271 | "cell_type": "markdown", |
|
272 | "metadata": {}, | |
|
243 | 273 | "source": [ |
|
244 | "In scikit-learn, an estimator is just a plain Python class that", | |
|
274 | "In scikit-learn, an estimator is just a plain Python class that\n", | |
|
245 | 275 | "implements the methods fit(X, Y) and predict(T)." |
|
246 | 276 | ] |
|
247 | 277 | }, |
|
248 | 278 | { |
|
249 | 279 | "cell_type": "markdown", |
|
280 | "metadata": {}, | |
|
250 | 281 | "source": [ |
|
251 | "An example of estimator is the class sklearn.svm.SVC that", | |
|
252 | "implements Support Vector Classification. The", | |
|
253 | "constructor of an estimator takes as arguments the parameters of the", | |
|
254 | "model, but for the time being, we will consider the estimator as a black", | |
|
282 | "An example of estimator is the class sklearn.svm.SVC that\n", | |
|
283 | "implements Support Vector Classification. The\n", | |
|
284 | "constructor of an estimator takes as arguments the parameters of the\n", | |
|
285 | "model, but for the time being, we will consider the estimator as a black\n", | |
|
255 | 286 | "box:" |
|
256 | 287 | ] |
|
257 | 288 | }, |
@@ -259,35 +290,39 b'' | |||
|
259 | 290 | "cell_type": "code", |
|
260 | 291 | "collapsed": false, |
|
261 | 292 | "input": [ |
|
262 | "from sklearn import svm", | |
|
293 | "from sklearn import svm\n", | |
|
263 | 294 | "clf = svm.SVC(gamma=0.001, C=100.)" |
|
264 | 295 | ], |
|
265 | 296 | "language": "python", |
|
297 | "metadata": {}, | |
|
266 | 298 | "outputs": [] |
|
267 | 299 | }, |
|
268 | 300 | { |
|
269 | 301 | "cell_type": "heading", |
|
270 | 302 | "level": 2, |
|
303 | "metadata": {}, | |
|
271 | 304 | "source": [ |
|
272 | 305 | "Choosing the parameters of the model" |
|
273 | 306 | ] |
|
274 | 307 | }, |
|
275 | 308 | { |
|
276 | 309 | "cell_type": "markdown", |
|
310 | "metadata": {}, | |
|
277 | 311 | "source": [ |
|
278 | "In this example we set the value of gamma manually. It is possible", | |
|
279 | "to automatically find good values for the parameters by using tools", | |
|
280 | "such as :ref:`grid search <grid_search>` and :ref:`cross validation", | |
|
312 | "In this example we set the value of gamma manually. It is possible\n", | |
|
313 | "to automatically find good values for the parameters by using tools\n", | |
|
314 | "such as :ref:`grid search <grid_search>` and :ref:`cross validation\n", | |
|
281 | 315 | "<cross_validation>`." |
|
282 | 316 | ] |
|
283 | 317 | }, |
|
284 | 318 | { |
|
285 | 319 | "cell_type": "markdown", |
|
320 | "metadata": {}, | |
|
286 | 321 | "source": [ |
|
287 | "We call our estimator instance clf as it is a classifier. It now must", | |
|
288 | "be fitted to the model, that is, it must learn from the model. This is", | |
|
289 | "done by passing our training set to the fit method. As a training", | |
|
290 | "set, let us use all the images of our dataset apart from the last", | |
|
322 | "We call our estimator instance clf as it is a classifier. It now must\n", | |
|
323 | "be fitted to the model, that is, it must learn from the model. This is\n", | |
|
324 | "done by passing our training set to the fit method. As a training\n", | |
|
325 | "set, let us use all the images of our dataset apart from the last\n", | |
|
291 | 326 | "one:" |
|
292 | 327 | ] |
|
293 | 328 | }, |
@@ -298,13 +333,15 b'' | |||
|
298 | 333 | "clf.fit(digits.data[:-1], digits.target[:-1])" |
|
299 | 334 | ], |
|
300 | 335 | "language": "python", |
|
336 | "metadata": {}, | |
|
301 | 337 | "outputs": [] |
|
302 | 338 | }, |
|
303 | 339 | { |
|
304 | 340 | "cell_type": "markdown", |
|
341 | "metadata": {}, | |
|
305 | 342 | "source": [ |
|
306 | "Now you can predict new values, in particular, we can ask to the", | |
|
307 | "classifier what is the digit of our last image in the digits dataset,", | |
|
343 | "Now you can predict new values, in particular, we can ask to the\n", | |
|
344 | "classifier what is the digit of our last image in the digits dataset,\n", | |
|
308 | 345 | "which we have not used to train the classifier:" |
|
309 | 346 | ] |
|
310 | 347 | }, |
@@ -315,40 +352,46 b'' | |||
|
315 | 352 | "clf.predict(digits.data[-1])" |
|
316 | 353 | ], |
|
317 | 354 | "language": "python", |
|
355 | "metadata": {}, | |
|
318 | 356 | "outputs": [] |
|
319 | 357 | }, |
|
320 | 358 | { |
|
321 | 359 | "cell_type": "markdown", |
|
360 | "metadata": {}, | |
|
322 | 361 | "source": [ |
|
323 | 362 | "The corresponding image is the following:" |
|
324 | 363 | ] |
|
325 | 364 | }, |
|
326 | 365 | { |
|
327 | 366 | "cell_type": "markdown", |
|
367 | "metadata": {}, | |
|
328 | 368 | "source": [ |
|
329 | "As you can see, it is a challenging task: the images are of poor", | |
|
369 | "As you can see, it is a challenging task: the images are of poor\n", | |
|
330 | 370 | "resolution. Do you agree with the classifier?" |
|
331 | 371 | ] |
|
332 | 372 | }, |
|
333 | 373 | { |
|
334 | 374 | "cell_type": "markdown", |
|
375 | "metadata": {}, | |
|
335 | 376 | "source": [ |
|
336 | "A complete example of this classification problem is available as an", | |
|
337 | "example that you can run and study:", | |
|
377 | "A complete example of this classification problem is available as an\n", | |
|
378 | "example that you can run and study:\n", | |
|
338 | 379 | ":ref:`example_plot_digits_classification.py`." |
|
339 | 380 | ] |
|
340 | 381 | }, |
|
341 | 382 | { |
|
342 | 383 | "cell_type": "heading", |
|
343 | 384 | "level": 2, |
|
385 | "metadata": {}, | |
|
344 | 386 | "source": [ |
|
345 | 387 | "Model persistence" |
|
346 | 388 | ] |
|
347 | 389 | }, |
|
348 | 390 | { |
|
349 | 391 | "cell_type": "markdown", |
|
392 | "metadata": {}, | |
|
350 | 393 | "source": [ |
|
351 | "It is possible to save a model in the scikit by using Python's built-in", | |
|
394 | "It is possible to save a model in the scikit by using Python's built-in\n", | |
|
352 | 395 | "persistence model, namely pickle:" |
|
353 | 396 | ] |
|
354 | 397 | }, |
@@ -356,27 +399,29 b'' | |||
|
356 | 399 | "cell_type": "code", |
|
357 | 400 | "collapsed": false, |
|
358 | 401 | "input": [ |
|
359 | "from sklearn import svm", | |
|
360 | "from sklearn import datasets", | |
|
361 | "clf = svm.SVC()", | |
|
362 | "iris = datasets.load_iris()", | |
|
363 | "X, y = iris.data, iris.target", | |
|
364 | "clf.fit(X, y)", | |
|
365 | "import pickle", | |
|
366 | "s = pickle.dumps(clf)", | |
|
367 | "clf2 = pickle.loads(s)", | |
|
368 | "clf2.predict(X[0])", | |
|
402 | "from sklearn import svm\n", | |
|
403 | "from sklearn import datasets\n", | |
|
404 | "clf = svm.SVC()\n", | |
|
405 | "iris = datasets.load_iris()\n", | |
|
406 | "X, y = iris.data, iris.target\n", | |
|
407 | "clf.fit(X, y)\n", | |
|
408 | "import pickle\n", | |
|
409 | "s = pickle.dumps(clf)\n", | |
|
410 | "clf2 = pickle.loads(s)\n", | |
|
411 | "clf2.predict(X[0])\n", | |
|
369 | 412 | "y[0]" |
|
370 | 413 | ], |
|
371 | 414 | "language": "python", |
|
415 | "metadata": {}, | |
|
372 | 416 | "outputs": [] |
|
373 | 417 | }, |
|
374 | 418 | { |
|
375 | 419 | "cell_type": "markdown", |
|
420 | "metadata": {}, | |
|
376 | 421 | "source": [ |
|
377 | "In the specific case of the scikit, it may be more interesting to use", | |
|
378 | "joblib's replacement of pickle (joblib.dump & joblib.load),", | |
|
379 | "which is more efficient on big data, but can only pickle to the disk", | |
|
422 | "In the specific case of the scikit, it may be more interesting to use\n", | |
|
423 | "joblib's replacement of pickle (joblib.dump & joblib.load),\n", | |
|
424 | "which is more efficient on big data, but can only pickle to the disk\n", | |
|
380 | 425 | "and not to a string:" |
|
381 | 426 | ] |
|
382 | 427 | }, |
@@ -384,13 +429,15 b'' | |||
|
384 | 429 | "cell_type": "code", |
|
385 | 430 | "collapsed": false, |
|
386 | 431 | "input": [ |
|
387 | "from sklearn.externals import joblib", | |
|
432 | "from sklearn.externals import joblib\n", | |
|
388 | 433 | "joblib.dump(clf, 'filename.pkl') # doctest: +SKIP" |
|
389 | 434 | ], |
|
390 | 435 | "language": "python", |
|
436 | "metadata": {}, | |
|
391 | 437 | "outputs": [] |
|
392 | 438 | } |
|
393 | ] | |
|
439 | ], | |
|
440 | "metadata": {} | |
|
394 | 441 | } |
|
395 | 442 | ] |
|
396 | 443 | } No newline at end of file |
General Comments 0
You need to be logged in to leave comments.
Login now