/automl-gfluz

Primary LanguagePythonMIT LicenseMIT

AUTOML

This is a library developed to incorporate useful properties and methods in relevant data science packages, such as scikit-learn and pycaret, in order to provide a pipeline which suits every supervised problem. Therefore, data scientists can spend less time working on building pipelines and use this time more wisely to create new features and tune the best model.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

from sklearn.model_selection import train_test_split

1 - Classification

df = pd.read_csv("juice.csv")
df.head()
Id Purchase WeekofPurchase StoreID PriceCH PriceMM DiscCH DiscMM SpecialCH SpecialMM LoyalCH SalePriceMM SalePriceCH PriceDiff Store7 PctDiscMM PctDiscCH ListPriceDiff STORE
0 1 CH 237 1 1.75 1.99 0.00 0.0 0 0 0.500000 1.99 1.75 0.24 No 0.000000 0.000000 0.24 1
1 2 CH 239 1 1.75 1.99 0.00 0.3 0 1 0.600000 1.69 1.75 -0.06 No 0.150754 0.000000 0.24 1
2 3 CH 245 1 1.86 2.09 0.17 0.0 0 0 0.680000 2.09 1.69 0.40 No 0.000000 0.091398 0.23 1
3 4 MM 227 1 1.69 1.69 0.00 0.0 0 0 0.400000 1.69 1.69 0.00 No 0.000000 0.000000 0.00 1
4 5 CH 228 7 1.69 1.69 0.00 0.0 0 0 0.956535 1.69 1.69 0.00 Yes 0.000000 0.000000 0.00 0 
X = df.iloc[:, 2:]
y = df.iloc[:, 1].apply(lambda x: 1 if x=="CH" else 0)
train_X, test_X, train_y, test_y = train_test_split(X, y, test_size=0.2, random_state=0)

Importing our class for classification problems

from automl.classification import ClassifierPyCaret

Defining object

my_clf = ClassifierPyCaret(metric="AUC")

Training all models and selecting the one with best performance

my_clf.fit(train_X, train_y)
Model Accuracy AUC Recall Prec. F1 Kappa
0 Linear Discriminant Analysis 0.8312 0.8994 0.8724 0.8589 0.8649 0.6396
1 Logistic Regression 0.8338 0.8965 0.8787 0.858 0.8674 0.6444
2 CatBoost Classifier 0.8104 0.889 0.8514 0.8459 0.8475 0.5963
3 Extreme Gradient Boosting 0.8078 0.8838 0.8599 0.8379 0.8475 0.5873
4 Gradient Boosting Classifier 0.8039 0.8804 0.8577 0.8338 0.8444 0.5789
5 Light Gradient Boosting Machine 0.7831 0.8745 0.8284 0.8255 0.8256 0.5381
6 Ada Boost Classifier 0.8169 0.8628 0.8787 0.8358 0.8563 0.6042
7 Random Forest Classifier 0.7831 0.8401 0.7951 0.848 0.8185 0.5493
8 K Neighbors Classifier 0.7818 0.8367 0.8451 0.8117 0.8278 0.5302
9 Naive Bayes 0.761 0.8308 0.7658 0.8381 0.7994 0.5049
10 Extra Trees Classifier 0.7532 0.8279 0.8034 0.8004 0.8006 0.4764
11 Quadratic Discriminant Analysis 0.713 0.7364 0.7844 0.7625 0.7716 0.3838
12 Decision Tree Classifier 0.7325 0.7222 0.7659 0.7991 0.7797 0.438
13 SVM - Linear Kernel 0.7727 0 0.8682 0.7921 0.8261 0.4988
14 Ridge Classifier 0.8338 0 0.8704 0.8641 0.8661 0.6465 
my_clf.best_model
LinearDiscriminantAnalysis(n_components=None, priors=None, shrinkage=0.001,
                           solver='lsqr', store_covariance=False, tol=0.0001)

Predicting

my_clf.predict(test_X)
array([0, 1, 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1,
       0, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0,
       0, 0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1,
       0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1,
       0, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 0,
       0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, 0,
       1, 1, 1, 0, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 1,
       0, 1, 1, 0, 1, 1, 1, 0, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1,
       1, 1, 0, 0, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0, 1, 0, 1, 1,
       1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1])

test_y.values
array([0, 1, 1, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0,
       0, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 1, 0,
       0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 0,
       0, 1, 0, 1, 0, 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 1,
       0, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1,
       0, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 1, 0, 1, 1, 0, 0, 0,
       1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 1,
       0, 0, 1, 0, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 1, 0, 1, 0, 1,
       1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 0,
       1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1])

Predicting score

my_clf.predict_proba(test_X)[:15]
array([[0.81394134, 0.18605866],
       [0.29250523, 0.70749477],
       [0.0556436 , 0.9443564 ],
       [0.34100222, 0.65899778],
       [0.30329665, 0.69670335],
       [0.93766687, 0.06233313],
       [0.79220565, 0.20779435],
       [0.20849993, 0.79150007],
       [0.03889261, 0.96110739],
       [0.96823484, 0.03176516],
       [0.03828661, 0.96171339],
       [0.05635964, 0.94364036],
       [0.34055076, 0.65944924],
       [0.01090129, 0.98909871],
       [0.00782886, 0.99217114]])

Preprocessed test set

my_clf.preprocess(test_X)
WeekofPurchase StoreID PriceCH PriceMM SpecialCH SpecialMM LoyalCH SalePriceMM SalePriceCH PriceDiff PctDiscMM PctDiscCH ListPriceDiff STORE Store7_Yes
928 -1.332854 -0.420137 -0.778278 0.026205 -0.411581 -0.436512 -1.199095 0.497515 -0.176508 0.562370 -0.581691 -0.446505 0.764189 0.958208 0.0
780 -1.462320 -1.284649 -1.769803 -0.724024 2.429655 -0.436512 0.038105 0.097597 -0.875681 0.562370 -0.581691 -0.446505 0.764189 -0.435549 0.0
564 0.738609 1.308888 -0.084210 0.326297 2.429655 -0.436512 0.082491 0.657482 -2.274027 1.840059 -0.581691 2.722738 0.484560 -1.132427 1.0
520 0.285476 -0.852393 -0.084210 0.701412 -0.411581 2.290890 0.341921 -0.742231 0.312913 -0.865637 1.253504 -0.446505 0.950608 0.261329 0.0
399 1.062275 1.308888 -0.084210 0.326297 -0.411581 -0.436512 -0.470514 0.657482 0.312913 0.449632 -0.581691 -0.446505 0.484560 -1.132427 1.0
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
231 -1.721253 1.308888 -1.769803 -2.974714 -0.411581 -0.436512 -0.127836 -1.102158 -0.875681 -0.565004 -0.581691 -0.446505 -2.032101 -1.132427 1.0
669 -0.038190 0.012119 1.204772 1.526665 2.429655 -0.436512 -0.241797 1.297351 1.221837 0.562370 -0.581691 -0.446505 0.764189 1.655086 0.0
821 -1.721253 0.012119 -0.778278 -2.224484 -0.411581 -0.436512 0.537512 -0.702239 -0.176508 -0.565004 -0.581691 -0.446505 -2.032101 1.655086 0.0
54 0.091277 0.012119 1.204772 1.526665 -0.411581 -0.436512 0.302929 1.297351 1.221837 0.562370 -0.581691 -0.446505 0.764189 1.655086 0.0
34 -0.232389 -0.420137 1.204772 1.076527 -0.411581 -0.436512 -0.099110 1.057400 1.221837 0.336895 -0.581691 -0.446505 0.204931 0.958208 0.0

214 rows × 15 columns

Evaluating final model

my_clf.evaluate(test_X, test_y)
Accuracy Recall Precision F1-Score AUC-ROC
Linear Discriminant Analysis 0.82243 0.844262 0.844262 0.844262 0.896605

Ploting curves to evaluate performace

my_clf.binary_evaluation_plot(test_X, test_y)

png

2 - Regression

df = pd.read_csv("boston.csv")
df.head()
crim zn indus chas nox rm age dis rad tax ptratio black lstat medv
0 0.00632 18.0 2.31 0 0.538 6.575 65.2 4.0900 1 296 15.3 396.90 4.98 24.0
1 0.02731 0.0 7.07 0 0.469 6.421 78.9 4.9671 2 242 17.8 396.90 9.14 21.6
2 0.02729 0.0 7.07 0 0.469 7.185 61.1 4.9671 2 242 17.8 392.83 4.03 34.7
3 0.03237 0.0 2.18 0 0.458 6.998 45.8 6.0622 3 222 18.7 394.63 2.94 33.4
4 0.06905 0.0 2.18 0 0.458 7.147 54.2 6.0622 3 222 18.7 396.90 5.33 36.2 
X = df.iloc[:, :-1]
y = df.iloc[:, -1]
train_X, test_X, train_y, test_y = train_test_split(X, y, test_size=0.2, random_state=0)

Importing our class for classification problems

from automl.regression import RegressorPyCaret

Defining object

my_reg = RegressorPyCaret(metric="RMSE")

Training all models and selecting the one with best performance

my_reg.fit(train_X, train_y)
Model MAE MSE RMSE R2 RMSLE MAPE
0 Extra Trees Regressor 2.0117 9.8317 3.0117 0.882 0.1313 0.1
1 CatBoost Regressor 2.0328 10.0328 3.0457 0.8825 0.1337 0.1004
2 Gradient Boosting Regressor 2.0817 10.3726 3.1033 0.8773 0.1392 0.1055
3 Extreme Gradient Boosting 2.1903 11.3859 3.2391 0.8657 0.1424 0.1093
4 Light Gradient Boosting Machine 2.2946 12.4216 3.429 0.8533 0.1456 0.1116
5 Random Forest 2.2884 12.7449 3.4698 0.8493 0.1494 0.1139
6 AdaBoost Regressor 2.6787 14.0651 3.6835 0.8322 0.1674 0.1377
7 K Neighbors Regressor 2.7906 19.2193 4.2479 0.777 0.1706 0.1285
8 Bayesian Ridge 3.1667 21.2922 4.546 0.7493 0.2214 0.1552
9 Ridge Regression 3.1846 21.357 4.5549 0.7483 0.2214 0.1558
10 Linear Regression 3.1907 21.3876 4.5588 0.7479 0.2215 0.156
11 Huber Regressor 3.02 21.9739 4.5865 0.7425 0.2423 0.1459
12 Random Sample Consensus 3.141 23.5192 4.6983 0.7248 0.257 0.1529
13 Decision Tree 3.1003 23.0186 4.7735 0.7244 0.2072 0.156
14 Least Angle Regression 3.4683 26.3764 5.0549 0.6804 0.2535 0.1757
15 Lasso Regression 3.586 26.3971 5.0742 0.6933 0.2454 0.1818
16 Elastic Net 3.6377 27.6603 5.2005 0.6777 0.2227 0.1786
17 Support Vector Machine 3.2355 29.4341 5.3048 0.6639 0.2138 0.1569
18 TheilSen Regressor 3.6964 33.5251 5.6568 0.6107 0.3268 0.1814
19 Passive Aggressive Regressor 4.3705 38.8214 6.203 0.5339 0.3688 0.2342
20 Orthogonal Matching Pursuit 4.7336 43.4094 6.4761 0.4999 0.3263 0.2297
21 Lasso Least Angle Regression 6.8136 86.403 9.261 -0.0147 0.3952 0.3674 
my_reg.best_model
ExtraTreesRegressor(bootstrap=False, ccp_alpha=0.0, criterion='mse',
                    max_depth=50, max_features='auto', max_leaf_nodes=None,
                    max_samples=None, min_impurity_decrease=0.0,
                    min_impurity_split=None, min_samples_leaf=1,
                    min_samples_split=7, min_weight_fraction_leaf=0.0,
                    n_estimators=230, n_jobs=None, oob_score=False,
                    random_state=4079, verbose=0, warm_start=False)

Predicting

my_reg.predict(test_X)
array([23.59301449, 23.86976812, 22.9758913 , 10.77536232, 21.7523913 ,
       20.45304348, 21.64648551, 20.05999275, 20.34348551, 19.65515942,
        8.15554348, 15.19625362, 15.24094203,  8.6743913 , 48.79373913,
       34.54835507, 21.51931159, 36.47942754, 25.99400725, 21.06105072,
       23.53934783, 22.12610145, 19.5293913 , 24.70913043, 20.46073188,
       17.92857246, 17.57008696, 16.32171014, 43.31463043, 19.55130435,
       15.54263768, 18.10597101, 21.12585507, 21.45034058, 24.27658696,
       19.78678986,  9.05228261, 25.52650725, 14.11785507, 15.47104348,
       22.96647826, 20.59884058, 21.83624638, 16.47018841, 24.62587681,
       23.3175942 , 20.49226087, 18.62208696, 15.1505942 , 24.18486232,
       16.4717971 , 20.65637681, 21.94365217, 38.90871739, 15.64312319,
       21.3097029 , 19.55873913, 19.6977971 , 19.44619565, 20.55236957,
       21.44344203, 22.00573188, 32.4110942 , 28.30823188, 19.36657246,
       27.20982609, 15.82344928, 22.58002899, 15.18563768, 22.05265217,
       20.84502174, 22.77226812, 24.42430435, 31.16600725, 26.7797029 ,
        8.60797101, 42.23894928, 22.59955797, 23.76918841, 19.81447826,
       25.12301449, 18.41964493, 19.59877536, 42.00841304, 41.19852899,
       23.88085507, 22.77155797, 14.20571739, 26.48037681, 15.24131884,
       19.44444203, 11.67923188, 22.47393478, 31.38963043, 21.19071739,
       22.03372464, 12.33304348, 23.81121014, 14.16386957, 18.81552899,
       23.94378986, 19.97833333])

test_y.values
array([22.6, 50. , 23. ,  8.3, 21.2, 19.9, 20.6, 18.7, 16.1, 18.6,  8.8,
       17.2, 14.9, 10.5, 50. , 29. , 23. , 33.3, 29.4, 21. , 23.8, 19.1,
       20.4, 29.1, 19.3, 23.1, 19.6, 19.4, 38.7, 18.7, 14.6, 20. , 20.5,
       20.1, 23.6, 16.8,  5.6, 50. , 14.5, 13.3, 23.9, 20. , 19.8, 13.8,
       16.5, 21.6, 20.3, 17. , 11.8, 27.5, 15.6, 23.1, 24.3, 42.8, 15.6,
       21.7, 17.1, 17.2, 15. , 21.7, 18.6, 21. , 33.1, 31.5, 20.1, 29.8,
       15.2, 15. , 27.5, 22.6, 20. , 21.4, 23.5, 31.2, 23.7,  7.4, 48.3,
       24.4, 22.6, 18.3, 23.3, 17.1, 27.9, 44.8, 50. , 23. , 21.4, 10.2,
       23.3, 23.2, 18.9, 13.4, 21.9, 24.8, 11.9, 24.3, 13.8, 24.7, 14.1,
       18.7, 28.1, 19.8])

Preprocessed test set

my_reg.preprocess(test_X)
crim zn indus chas nox rm age dis rad tax ptratio black lstat
329 -0.408359 -0.499608 -1.128729 -0.272888 -0.833369 0.044972 -1.846215 0.695069 -0.624648 0.159137 -0.712729 0.185476 -0.736103
371 0.719251 -0.499608 0.998884 -0.272888 0.652840 -0.123657 1.103327 -1.251749 1.687378 1.542121 0.792674 0.083165 -0.435692
219 -0.402575 -0.499608 0.396108 3.664502 -0.051154 0.102623 0.832597 -0.195833 -0.509046 -0.743319 -0.940821 0.394727 -0.302632
403 2.634810 -0.499608 0.998884 -0.272888 1.191699 -1.373240 0.960837 -0.994916 1.687378 1.542121 0.792674 0.430412 0.968974
78 -0.409685 -0.499608 0.244340 -0.272888 -1.033268 -0.100597 -0.545994 0.598583 -0.509046 -0.028387 0.108400 0.311840 -0.050232
... ... ... ... ... ... ... ... ... ... ... ... ... ...
56 -0.414103 3.100922 -1.486672 -0.272888 -1.267933 0.117036 -1.187199 2.606998 -0.855850 -0.526496 -0.530256 0.430412 -0.951467
455 0.168069 -0.499608 0.998884 -0.272888 1.365525 0.321696 0.622424 -0.642174 1.687378 1.542121 0.792674 -3.476599 0.744008
60 -0.398260 0.559372 -0.858124 -0.272888 -0.894208 -0.808261 -0.100713 1.662728 -0.162242 -0.696439 0.564583 0.410198 0.060880
213 -0.399343 -0.499608 -0.076377 -0.272888 -0.581322 0.105505 -1.308316 0.084292 -0.624648 -0.737459 0.062782 0.305177 -0.456268
108 -0.400881 -0.499608 -0.367026 -0.272888 -0.311892 0.248191 1.000022 -0.643570 -0.509046 -0.110428 1.112002 0.411666 -0.059834

102 rows × 13 columns

Evaluating final model

my_reg.evaluate(test_X, test_y)
MAE MSE RMSE R2 RMLSE MAPE
Extra Trees Regressor 2.804562 23.271022 4.824005 0.714215 0.182163 0.128135

Checking on residuals

my_reg.plot_analysis(test_X, test_y)

png

Copyright (c) 2018 The Python Packaging Authority

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.