Machine Learning Programming Workshop

3.1 Implementation with Sci-Kit Learn

Prepared By: Cheong Shiu Hong (FTFNCE)


Contents

Demonstrate with Digit Dataset, then let students play around with faces dataset, feel free to import other algorithms from sklearn to try and compare accuracy.


In [1]:
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import sklearn as sk
from sklearn import datasets
In [2]:
%matplotlib inline
In [3]:
from sklearn.linear_model import LogisticRegression
from sklearn.neural_network import MLPClassifier


1) Digits Dataset

return to top

Load Digits Dataset from Sklearn

In [4]:
digits = datasets.load_digits()

Explore Dataset

In [5]:
digits.keys()
Out[5]:
dict_keys(['data', 'target', 'target_names', 'images', 'DESCR'])
In [6]:
digits['images'].shape, digits['data'].shape, digits['target'].shape
Out[6]:
((1797, 8, 8), (1797, 64), (1797,))

Visualize Images

In [7]:
fig=plt.figure(figsize=(10, 4))
columns = 5
rows = 2

for i in range(columns*rows):
    img = digits['images'][i]
    fig.add_subplot(rows, columns, i+1)
    plt.imshow(img)
    
plt.show()

Splitting the Dataset

In [8]:
X = digits['data'][:1600]
Y = digits['target'][:1600]
X_val = digits['data'][1600:]
Y_val = digits['target'][1600:]


Logistic Regression

In [9]:
logreg = LogisticRegression(solver='liblinear', max_iter=1000, multi_class='ovr')
In [10]:
logreg.fit(digits['data'], digits['target'])
Out[10]:
LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
          intercept_scaling=1, max_iter=1000, multi_class='ovr',
          n_jobs=None, penalty='l2', random_state=None, solver='liblinear',
          tol=0.0001, verbose=0, warm_start=False)
In [11]:
logreg.score(X, Y)
Out[11]:
0.99375
In [12]:
logreg.score(X_val, Y_val)
Out[12]:
0.9898477157360406


Multi-Layer Perceptron (Neural Network)

In [13]:
NN = MLPClassifier(max_iter=1000, hidden_layer_sizes=(16,32,16))
In [14]:
NN.fit(X, Y)
Out[14]:
MLPClassifier(activation='relu', alpha=0.0001, batch_size='auto', beta_1=0.9,
       beta_2=0.999, early_stopping=False, epsilon=1e-08,
       hidden_layer_sizes=(16, 32, 16), learning_rate='constant',
       learning_rate_init=0.001, max_iter=1000, momentum=0.9,
       n_iter_no_change=10, nesterovs_momentum=True, power_t=0.5,
       random_state=None, shuffle=True, solver='adam', tol=0.0001,
       validation_fraction=0.1, verbose=False, warm_start=False)
In [15]:
NN.score(X, Y)
Out[15]:
1.0
In [16]:
NN.score(X_val, Y_val)
Out[16]:
0.9187817258883249


2) Faces Dataset

return to top

Load Faces Dataset from Sklearn

In [17]:
faces = datasets.fetch_olivetti_faces()

downloading Olivetti faces from https://ndownloader.figshare.com/files/5976027 to C:\Users\cheon\scikit_learn_data

Explore Dataset

In [18]:
faces.keys()
Out[18]:
dict_keys(['data', 'images', 'target', 'DESCR'])
In [19]:
faces['images'].shape, faces['data'].shape, faces['target'].shape
Out[19]:
((400, 64, 64), (400, 4096), (400,))

Shuffling Dataset

In [20]:
s = np.arange(len(faces['data']))
np.random.shuffle(s)
In [21]:
data = faces['data'][s]
images = faces['images'][s]
target = faces['target'][s]

Visualizing Images

In [22]:
fig=plt.figure(figsize=(10, 8))
columns = 5
rows = 4

for i in range(columns*rows):
    img = images[i]
    fig.add_subplot(rows, columns, i+1)
    plt.imshow(img)
    
plt.show()

Splitting the Dataset

In [23]:
from sklearn.model_selection import train_test_split

X, X_val, Y, Y_val = train_test_split(data, target, test_size=0.15, shuffle=True)


Logistic Regression

In [24]:
logreg = LogisticRegression(solver='liblinear', multi_class='ovr', verbose=1)
In [25]:
logreg.fit(X, Y)

[LibLinear]
Out[25]:
LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
          intercept_scaling=1, max_iter=100, multi_class='ovr',
          n_jobs=None, penalty='l2', random_state=None, solver='liblinear',
          tol=0.0001, verbose=1, warm_start=False)
In [26]:
logreg.score(X, Y)
Out[26]:
1.0
In [27]:
logreg.score(X_val, Y_val)
Out[27]:
0.8833333333333333


Neural Network

In [28]:
NN = MLPClassifier(max_iter=1000, hidden_layer_sizes=(64,128,32), verbose=1)
In [29]:
NN.fit(X, Y)

Iteration 1, loss = 3.75954071
Iteration 2, loss = 3.72140950
Iteration 3, loss = 3.70619785
Iteration 4, loss = 3.68376290
Iteration 5, loss = 3.68153736
Iteration 6, loss = 3.66720544
Iteration 7, loss = 3.65718644
Iteration 8, loss = 3.64696736
Iteration 9, loss = 3.63108768
Iteration 10, loss = 3.61720944
Iteration 11, loss = 3.61465325
Iteration 12, loss = 3.59060380
Iteration 13, loss = 3.57816035
Iteration 14, loss = 3.55691958
Iteration 15, loss = 3.54625345
Iteration 16, loss = 3.53181864
Iteration 17, loss = 3.51005005
Iteration 18, loss = 3.49752272
Iteration 19, loss = 3.48133042
Iteration 20, loss = 3.46167566
Iteration 21, loss = 3.43496562
Iteration 22, loss = 3.41331253
Iteration 23, loss = 3.38550214
Iteration 24, loss = 3.36396797
Iteration 25, loss = 3.34827182
Iteration 26, loss = 3.32193922
Iteration 27, loss = 3.29446621
Iteration 28, loss = 3.27982908
Iteration 29, loss = 3.25549940
Iteration 30, loss = 3.24871105
Iteration 31, loss = 3.23136084
Iteration 32, loss = 3.19844415
Iteration 33, loss = 3.17389748
Iteration 34, loss = 3.13645985
Iteration 35, loss = 3.11111321
Iteration 36, loss = 3.07590951
Iteration 37, loss = 3.05748153
Iteration 38, loss = 3.00648661
Iteration 39, loss = 2.97882342
Iteration 40, loss = 2.95222317
Iteration 41, loss = 2.92433482
Iteration 42, loss = 2.87766698
Iteration 43, loss = 2.84157559
Iteration 44, loss = 2.80229867
Iteration 45, loss = 2.77733917
Iteration 46, loss = 2.73528812
Iteration 47, loss = 2.68913375
Iteration 48, loss = 2.65344173
Iteration 49, loss = 2.61353237
Iteration 50, loss = 2.57332767
Iteration 51, loss = 2.53151929
Iteration 52, loss = 2.49047507
Iteration 53, loss = 2.44962179
Iteration 54, loss = 2.39516765
Iteration 55, loss = 2.35388779
Iteration 56, loss = 2.31397803
Iteration 57, loss = 2.26433314
Iteration 58, loss = 2.21760090
Iteration 59, loss = 2.16825672
Iteration 60, loss = 2.12607762
Iteration 61, loss = 2.07491906
Iteration 62, loss = 2.04225005
Iteration 63, loss = 1.99339672
Iteration 64, loss = 1.93721357
Iteration 65, loss = 1.89332128
Iteration 66, loss = 1.86803934
Iteration 67, loss = 1.79924067
Iteration 68, loss = 1.75339201
Iteration 69, loss = 1.70378525
Iteration 70, loss = 1.67736540
Iteration 71, loss = 1.64136275
Iteration 72, loss = 1.58711864
Iteration 73, loss = 1.54917226
Iteration 74, loss = 1.55913130
Iteration 75, loss = 1.49538858
Iteration 76, loss = 1.47076767
Iteration 77, loss = 1.42799106
Iteration 78, loss = 1.39227456
Iteration 79, loss = 1.32316739
Iteration 80, loss = 1.28982082
Iteration 81, loss = 1.25309132
Iteration 82, loss = 1.22602660
Iteration 83, loss = 1.20845928
Iteration 84, loss = 1.15448441
Iteration 85, loss = 1.14477938
Iteration 86, loss = 1.13661710
Iteration 87, loss = 1.09302458
Iteration 88, loss = 1.05183419
Iteration 89, loss = 1.02013175
Iteration 90, loss = 0.99427615
Iteration 91, loss = 0.97625919
Iteration 92, loss = 0.97179443
Iteration 93, loss = 0.96213442
Iteration 94, loss = 0.94417529
Iteration 95, loss = 0.89343697
Iteration 96, loss = 0.86868141
Iteration 97, loss = 0.83869770
Iteration 98, loss = 0.84988500
Iteration 99, loss = 0.81855619
Iteration 100, loss = 0.80149408
Iteration 101, loss = 0.78258410
Iteration 102, loss = 0.76940857
Iteration 103, loss = 0.74374014
Iteration 104, loss = 0.72535538
Iteration 105, loss = 0.69463459
Iteration 106, loss = 0.69320310
Iteration 107, loss = 0.65589089
Iteration 108, loss = 0.64180526
Iteration 109, loss = 0.62316153
Iteration 110, loss = 0.59844031
Iteration 111, loss = 0.59853173
Iteration 112, loss = 0.58465789
Iteration 113, loss = 0.57435614
Iteration 114, loss = 0.56047892
Iteration 115, loss = 0.55440207
Iteration 116, loss = 0.53871299
Iteration 117, loss = 0.53495423
Iteration 118, loss = 0.51459883
Iteration 119, loss = 0.50563233
Iteration 120, loss = 0.48999512
Iteration 121, loss = 0.47660723
Iteration 122, loss = 0.47631254
Iteration 123, loss = 0.46928231
Iteration 124, loss = 0.44455376
Iteration 125, loss = 0.44597476
Iteration 126, loss = 0.43617797
Iteration 127, loss = 0.43218209
Iteration 128, loss = 0.43067635
Iteration 129, loss = 0.44426868
Iteration 130, loss = 0.41005727
Iteration 131, loss = 0.41884195
Iteration 132, loss = 0.43447502
Iteration 133, loss = 0.41371142
Iteration 134, loss = 0.37829587
Iteration 135, loss = 0.38682798
Iteration 136, loss = 0.38506825
Iteration 137, loss = 0.36213362
Iteration 138, loss = 0.34255904
Iteration 139, loss = 0.34609381
Iteration 140, loss = 0.33686154
Iteration 141, loss = 0.31842860
Iteration 142, loss = 0.32875350
Iteration 143, loss = 0.31977963
Iteration 144, loss = 0.30814497
Iteration 145, loss = 0.30265359
Iteration 146, loss = 0.29568893
Iteration 147, loss = 0.29081939
Iteration 148, loss = 0.29069644
Iteration 149, loss = 0.27925275
Iteration 150, loss = 0.27504099
Iteration 151, loss = 0.27051837
Iteration 152, loss = 0.26244727
Iteration 153, loss = 0.25657315
Iteration 154, loss = 0.25383056
Iteration 155, loss = 0.24638926
Iteration 156, loss = 0.24002387
Iteration 157, loss = 0.24658428
Iteration 158, loss = 0.23892174
Iteration 159, loss = 0.23154157
Iteration 160, loss = 0.22787429
Iteration 161, loss = 0.24497651
Iteration 162, loss = 0.23156868
Iteration 163, loss = 0.22120913
Iteration 164, loss = 0.22866860
Iteration 165, loss = 0.22689475
Iteration 166, loss = 0.21326342
Iteration 167, loss = 0.20711492
Iteration 168, loss = 0.21083784
Iteration 169, loss = 0.19768966
Iteration 170, loss = 0.19196966
Iteration 171, loss = 0.19025933
Iteration 172, loss = 0.18947794
Iteration 173, loss = 0.18634857
Iteration 174, loss = 0.18300351
Iteration 175, loss = 0.18574889
Iteration 176, loss = 0.17894688
Iteration 177, loss = 0.17840593
Iteration 178, loss = 0.16738274
Iteration 179, loss = 0.16278082
Iteration 180, loss = 0.16525237
Iteration 181, loss = 0.15945735
Iteration 182, loss = 0.15338608
Iteration 183, loss = 0.15013074
Iteration 184, loss = 0.14846023
Iteration 185, loss = 0.14362081
Iteration 186, loss = 0.14464999
Iteration 187, loss = 0.14179348
Iteration 188, loss = 0.14007166
Iteration 189, loss = 0.13683656
Iteration 190, loss = 0.13529510
Iteration 191, loss = 0.13184997
Iteration 192, loss = 0.13319329
Iteration 193, loss = 0.12867006
Iteration 194, loss = 0.12691674
Iteration 195, loss = 0.12388174
Iteration 196, loss = 0.12129520
Iteration 197, loss = 0.12144853
Iteration 198, loss = 0.11742077
Iteration 199, loss = 0.11685327
Iteration 200, loss = 0.11659604
Iteration 201, loss = 0.11212597
Iteration 202, loss = 0.11238415
Iteration 203, loss = 0.10977778
Iteration 204, loss = 0.10625456
Iteration 205, loss = 0.10500364
Iteration 206, loss = 0.10397974
Iteration 207, loss = 0.10138307
Iteration 208, loss = 0.10068983
Iteration 209, loss = 0.10044981
Iteration 210, loss = 0.09820354
Iteration 211, loss = 0.09681749
Iteration 212, loss = 0.09594799
Iteration 213, loss = 0.09514420
Iteration 214, loss = 0.09519489
Iteration 215, loss = 0.09213506
Iteration 216, loss = 0.09046436
Iteration 217, loss = 0.09004049
Iteration 218, loss = 0.08985042
Iteration 219, loss = 0.08777924
Iteration 220, loss = 0.08503973
Iteration 221, loss = 0.08696804
Iteration 222, loss = 0.08305980
Iteration 223, loss = 0.08518691
Iteration 224, loss = 0.08043955
Iteration 225, loss = 0.07977641
Iteration 226, loss = 0.07986072
Iteration 227, loss = 0.08040168
Iteration 228, loss = 0.07708025
Iteration 229, loss = 0.07505198
Iteration 230, loss = 0.07602573
Iteration 231, loss = 0.07276219
Iteration 232, loss = 0.07562113
Iteration 233, loss = 0.07114271
Iteration 234, loss = 0.06978438
Iteration 235, loss = 0.06880738
Iteration 236, loss = 0.06996384
Iteration 237, loss = 0.06570266
Iteration 238, loss = 0.06669985
Iteration 239, loss = 0.06482239
Iteration 240, loss = 0.06402789
Iteration 241, loss = 0.06531135
Iteration 242, loss = 0.06280282
Iteration 243, loss = 0.06236728
Iteration 244, loss = 0.06286855
Iteration 245, loss = 0.06056425
Iteration 246, loss = 0.06014335
Iteration 247, loss = 0.05853573
Iteration 248, loss = 0.05723045
Iteration 249, loss = 0.05739741
Iteration 250, loss = 0.05481132
Iteration 251, loss = 0.05721661
Iteration 252, loss = 0.05524616
Iteration 253, loss = 0.05538381
Iteration 254, loss = 0.05325890
Iteration 255, loss = 0.05295593
Iteration 256, loss = 0.05275334
Iteration 257, loss = 0.04955006
Iteration 258, loss = 0.05139339
Iteration 259, loss = 0.04921451
Iteration 260, loss = 0.04931748
Iteration 261, loss = 0.04837648
Iteration 262, loss = 0.04710878
Iteration 263, loss = 0.04723575
Iteration 264, loss = 0.04614887
Iteration 265, loss = 0.04621279
Iteration 266, loss = 0.04605704
Iteration 267, loss = 0.04540013
Iteration 268, loss = 0.04544381
Iteration 269, loss = 0.04328662
Iteration 270, loss = 0.04382627
Iteration 271, loss = 0.04234309
Iteration 272, loss = 0.04218345
Iteration 273, loss = 0.04129309
Iteration 274, loss = 0.04030642
Iteration 275, loss = 0.04055009
Iteration 276, loss = 0.03951972
Iteration 277, loss = 0.03992980
Iteration 278, loss = 0.03867121
Iteration 279, loss = 0.03785469
Iteration 280, loss = 0.03802675
Iteration 281, loss = 0.03737693
Iteration 282, loss = 0.03770338
Iteration 283, loss = 0.03622505
Iteration 284, loss = 0.03686695
Iteration 285, loss = 0.03556263
Iteration 286, loss = 0.03647389
Iteration 287, loss = 0.03474954
Iteration 288, loss = 0.03502431
Iteration 289, loss = 0.03475862
Iteration 290, loss = 0.03426351
Iteration 291, loss = 0.03363183
Iteration 292, loss = 0.03349951
Iteration 293, loss = 0.03277998
Iteration 294, loss = 0.03235946
Iteration 295, loss = 0.03203785
Iteration 296, loss = 0.03145067
Iteration 297, loss = 0.03155733
Iteration 298, loss = 0.03137658
Iteration 299, loss = 0.03134420
Iteration 300, loss = 0.03019447
Iteration 301, loss = 0.03042876
Iteration 302, loss = 0.02975197
Iteration 303, loss = 0.02924291
Iteration 304, loss = 0.02888459
Iteration 305, loss = 0.02880626
Iteration 306, loss = 0.02834854
Iteration 307, loss = 0.02822452
Iteration 308, loss = 0.02768527
Iteration 309, loss = 0.02764576
Iteration 310, loss = 0.02725293
Iteration 311, loss = 0.02739186
Iteration 312, loss = 0.02692737
Iteration 313, loss = 0.02696154
Iteration 314, loss = 0.02662775
Iteration 315, loss = 0.02618628
Iteration 316, loss = 0.02565593
Iteration 317, loss = 0.02552524
Iteration 318, loss = 0.02495794
Iteration 319, loss = 0.02504031
Iteration 320, loss = 0.02465111
Iteration 321, loss = 0.02433126
Iteration 322, loss = 0.02406670
Iteration 323, loss = 0.02404056
Iteration 324, loss = 0.02363011
Iteration 325, loss = 0.02374876
Iteration 326, loss = 0.02354046
Iteration 327, loss = 0.02302479
Iteration 328, loss = 0.02286175
Iteration 329, loss = 0.02257783
Iteration 330, loss = 0.02284992
Iteration 331, loss = 0.02230184
Iteration 332, loss = 0.02262902
Iteration 333, loss = 0.02168639
Iteration 334, loss = 0.02195591
Iteration 335, loss = 0.02171162
Iteration 336, loss = 0.02165383
Iteration 337, loss = 0.02109661
Iteration 338, loss = 0.02131012
Iteration 339, loss = 0.02080546
Iteration 340, loss = 0.02075935
Iteration 341, loss = 0.02038401
Iteration 342, loss = 0.02058243
Iteration 343, loss = 0.02002892
Iteration 344, loss = 0.02011800
Iteration 345, loss = 0.01960437
Iteration 346, loss = 0.01973390
Iteration 347, loss = 0.01905136
Iteration 348, loss = 0.01914744
Iteration 349, loss = 0.01889186
Iteration 350, loss = 0.01867411
Iteration 351, loss = 0.01844701
Iteration 352, loss = 0.01825469
Iteration 353, loss = 0.01851393
Iteration 354, loss = 0.01800076
Iteration 355, loss = 0.01821595
Iteration 356, loss = 0.01807276
Iteration 357, loss = 0.01756892
Iteration 358, loss = 0.01735994
Iteration 359, loss = 0.01731166
Iteration 360, loss = 0.01713034
Iteration 361, loss = 0.01706996
Iteration 362, loss = 0.01678749
Iteration 363, loss = 0.01676352
Iteration 364, loss = 0.01661132
Iteration 365, loss = 0.01648569
Iteration 366, loss = 0.01638358
Iteration 367, loss = 0.01618822
Iteration 368, loss = 0.01610310
Iteration 369, loss = 0.01581669
Iteration 370, loss = 0.01577435
Iteration 371, loss = 0.01570061
Iteration 372, loss = 0.01550728
Iteration 373, loss = 0.01542710
Iteration 374, loss = 0.01518061
Iteration 375, loss = 0.01526464
Iteration 376, loss = 0.01500148
Iteration 377, loss = 0.01494031
Iteration 378, loss = 0.01472426
Iteration 379, loss = 0.01488313
Iteration 380, loss = 0.01446523
Iteration 381, loss = 0.01444317
Iteration 382, loss = 0.01446955
Iteration 383, loss = 0.01422004
Iteration 384, loss = 0.01456266
Iteration 385, loss = 0.01400591
Iteration 386, loss = 0.01420237
Iteration 387, loss = 0.01378501
Iteration 388, loss = 0.01393369
Iteration 389, loss = 0.01378539
Iteration 390, loss = 0.01358504
Iteration 391, loss = 0.01339481
Iteration 392, loss = 0.01316152
Iteration 393, loss = 0.01331656
Iteration 394, loss = 0.01305214
Iteration 395, loss = 0.01313528
Iteration 396, loss = 0.01292357
Iteration 397, loss = 0.01281346
Iteration 398, loss = 0.01278512
Iteration 399, loss = 0.01266124
Iteration 400, loss = 0.01275489
Iteration 401, loss = 0.01246800
Iteration 402, loss = 0.01238718
Iteration 403, loss = 0.01222806
Iteration 404, loss = 0.01232714
Iteration 405, loss = 0.01211633
Iteration 406, loss = 0.01213327
Iteration 407, loss = 0.01192353
Iteration 408, loss = 0.01179859
Iteration 409, loss = 0.01181734
Iteration 410, loss = 0.01167429
Iteration 411, loss = 0.01161459
Iteration 412, loss = 0.01150635
Iteration 413, loss = 0.01137979
Iteration 414, loss = 0.01136293
Iteration 415, loss = 0.01124148
Iteration 416, loss = 0.01116909
Iteration 417, loss = 0.01106958
Iteration 418, loss = 0.01112885
Iteration 419, loss = 0.01102717
Iteration 420, loss = 0.01100092
Iteration 421, loss = 0.01088462
Iteration 422, loss = 0.01071617
Iteration 423, loss = 0.01067515
Iteration 424, loss = 0.01061826
Iteration 425, loss = 0.01058898
Iteration 426, loss = 0.01043970
Iteration 427, loss = 0.01046051
Iteration 428, loss = 0.01036475
Iteration 429, loss = 0.01025783
Iteration 430, loss = 0.01016333
Iteration 431, loss = 0.01011236
Iteration 432, loss = 0.01002891
Iteration 433, loss = 0.00997298
Iteration 434, loss = 0.00990440
Iteration 435, loss = 0.00987405
Iteration 436, loss = 0.00977368
Iteration 437, loss = 0.00972299
Iteration 438, loss = 0.00965307
Iteration 439, loss = 0.00963013
Iteration 440, loss = 0.00963374
Iteration 441, loss = 0.00947792
Iteration 442, loss = 0.00944945
Iteration 443, loss = 0.00936516
Iteration 444, loss = 0.00936413
Iteration 445, loss = 0.00933189
Iteration 446, loss = 0.00925145
Iteration 447, loss = 0.00915292
Iteration 448, loss = 0.00909979
Iteration 449, loss = 0.00908135
Iteration 450, loss = 0.00893894
Iteration 451, loss = 0.00897952
Iteration 452, loss = 0.00894545
Iteration 453, loss = 0.00875862
Iteration 454, loss = 0.00879139
Iteration 455, loss = 0.00872836
Iteration 456, loss = 0.00870548
Iteration 457, loss = 0.00859446
Iteration 458, loss = 0.00853730
Iteration 459, loss = 0.00852085
Iteration 460, loss = 0.00844686
Iteration 461, loss = 0.00843206
Iteration 462, loss = 0.00833343
Iteration 463, loss = 0.00836799
Iteration 464, loss = 0.00823136
Iteration 465, loss = 0.00823016
Iteration 466, loss = 0.00817902
Iteration 467, loss = 0.00808829
Iteration 468, loss = 0.00804319
Iteration 469, loss = 0.00799800
Iteration 470, loss = 0.00800092
Iteration 471, loss = 0.00787082
Iteration 472, loss = 0.00781751
Iteration 473, loss = 0.00785724
Iteration 474, loss = 0.00775055
Iteration 475, loss = 0.00772799
Iteration 476, loss = 0.00764483
Iteration 477, loss = 0.00769345
Iteration 478, loss = 0.00761416
Iteration 479, loss = 0.00753918
Iteration 480, loss = 0.00749739
Iteration 481, loss = 0.00748003
Iteration 482, loss = 0.00742645
Training loss did not improve more than tol=0.000100 for 10 consecutive epochs. Stopping.
Out[29]:
MLPClassifier(activation='relu', alpha=0.0001, batch_size='auto', beta_1=0.9,
       beta_2=0.999, early_stopping=False, epsilon=1e-08,
       hidden_layer_sizes=(64, 128, 32), learning_rate='constant',
       learning_rate_init=0.001, max_iter=1000, momentum=0.9,
       n_iter_no_change=10, nesterovs_momentum=True, power_t=0.5,
       random_state=None, shuffle=True, solver='adam', tol=0.0001,
       validation_fraction=0.1, verbose=1, warm_start=False)
In [30]:
NN.score(X, Y)
Out[30]:
1.0
In [31]:
NN.score(X_val, Y_val)
Out[31]:
0.75




Previous: Next: