!pip install cybooster --upgrade --no-cache-dir

from cybooster import BoosterClassifier, BoosterRegressor
from sklearn.datasets import load_iris, load_diabetes, load_breast_cancer, load_digits, load_wine
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, mean_squared_error, root_mean_squared_error
from sklearn.linear_model import LinearRegression
from time import time


# Regression Examples
X, y = load_diabetes(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
regressor = BoosterRegressor(obj=LinearRegression(), n_estimators=100, learning_rate=0.1,
                             n_hidden_features=10, verbose=1, seed=42)
start = time()
regressor.fit(X_train, y_train)
y_pred = regressor.predict(X_test)
print(f"Elapsed: {time() - start} s")
rmse = root_mean_squared_error(y_test, y_pred)
print(f"RMSE for regression: {rmse:.4f}")

# Classification Example
X, y = load_iris(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
classifier = BoosterClassifier(obj=LinearRegression(), n_estimators=100, learning_rate=0.1,
                               n_hidden_features=10, verbose=1, seed=42)
start = time()
try:
    classifier.fit(X_train, y_train)
except Exception as e: # this is for Windows users
    y_train = y_train.astype('int32')
    classifier.fit(X_train, y_train)
y_pred = classifier.predict(X_test)
print(f"Elapsed: {time() - start} s")
accuracy = accuracy_score(y_test, y_pred)
print(f"Accuracy for classification: {accuracy:.4f}")

X, y = load_wine(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
classifier = BoosterClassifier(obj=LinearRegression(), n_estimators=100, learning_rate=0.1,
                               n_hidden_features=10, verbose=1, seed=42)
start = time()
try:
    classifier.fit(X_train, y_train)
except Exception as e: # this is for Windows users
    y_train = y_train.astype('int32')
    classifier.fit(X_train, y_train)
y_pred = classifier.predict(X_test)
print(f"Elapsed: {time() - start} s")
accuracy = accuracy_score(y_test, y_pred)
print(f"Accuracy for classification: {accuracy:.4f}")

X, y = load_breast_cancer(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
classifier = BoosterClassifier(obj=LinearRegression(), n_estimators=100, learning_rate=0.1,
                               n_hidden_features=10, verbose=1, seed=42)
start = time()
try:
    classifier.fit(X_train, y_train)
except Exception as e:
    y_train = y_train.astype('int32')
    classifier.fit(X_train, y_train)
y_pred = classifier.predict(X_test)
print(f"Elapsed: {time() - start} s")
accuracy = accuracy_score(y_test, y_pred)
print(f"Accuracy for classification: {accuracy:.4f}")

X, y = load_digits(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
classifier = BoosterClassifier(obj=LinearRegression(), n_estimators=100, learning_rate=0.1,
                               n_hidden_features=10, verbose=1, seed=42)
start = time()
try:
    classifier.fit(X_train, y_train)
except Exception as e:
    y_train = y_train.astype('int32')
    classifier.fit(X_train, y_train)
y_pred = classifier.predict(X_test)
print(f"Elapsed: {time() - start} s")
accuracy = accuracy_score(y_test, y_pred)
print(f"Accuracy for classification: {accuracy:.4f}")

100%|██████████| 100/100 [00:00<00:00, 246.01it/s]

Elapsed: 0.47596025466918945 s
RMSE for regression: 50.0380

100%|██████████| 100/100 [00:00<00:00, 373.52it/s]

Elapsed: 0.3156728744506836 s
Accuracy for classification: 1.0000

100%|██████████| 100/100 [00:00<00:00, 314.06it/s]

Elapsed: 0.40644145011901855 s
Accuracy for classification: 1.0000

100%|██████████| 100/100 [00:00<00:00, 244.04it/s]

Elapsed: 0.47872471809387207 s
Accuracy for classification: 0.9649

100%|██████████| 100/100 [00:07<00:00, 12.72it/s]

Elapsed: 7.97456169128418 s
Accuracy for classification: 0.9500

import pandas as pd
from sklearn.utils import all_estimators
from tqdm import tqdm
from sklearn.utils.multiclass import type_of_target
from sklearn.datasets import fetch_california_housing


# Get all scikit-learn regressors
estimators = all_estimators(type_filter='regressor')

results_regressors = []
results_classifiers = []

verbose = 0

for name, RegressorClass in tqdm(estimators):

    if name in ['MultiOutputRegressor', 'MultiOutputClassifier', 'StackingRegressor', 'StackingClassifier',
                    'VotingRegressor', 'VotingClassifier', 'TransformedTargetRegressor', 'RegressorChain',
                    'GradientBoostingRegressor', 'HistGradientBoostingRegressor', 'RandomForestRegressor',
                    'ExtraTreesRegressor', 'MLPRegressor']:

        continue

    try:
      print(f"\nRunning with {name}")
      # Regression Examples
      X, y = load_diabetes(return_X_y=True)
      X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
      regressor = BoosterRegressor(obj=RegressorClass(), n_estimators=100, learning_rate=0.1,
                                  n_hidden_features=10, verbose=verbose, seed=42)
      start = time()
      regressor.fit(X_train, y_train)
      y_pred = regressor.predict(X_test)
      print(f"Elapsed: {time() - start} s")
      rmse = root_mean_squared_error(y_test, y_pred)
      print(f"RMSE for regression: {rmse:.4f}")
      results_regressors.append(["diabetes", name, rmse])

      X, y = fetch_california_housing(return_X_y=True)
      X_train, X_test, y_train, y_test = train_test_split(X[:1000,:], y[:1000], test_size=0.2, random_state=42)
      regressor = BoosterRegressor(obj=RegressorClass(), n_estimators=100, learning_rate=0.1,
                                  n_hidden_features=10, verbose=verbose, seed=42)
      start = time()
      regressor.fit(X_train, y_train)
      y_pred = regressor.predict(X_test)
      print(f"Elapsed: {time() - start} s")
      rmse = root_mean_squared_error(y_test, y_pred)
      print(f"RMSE for regression: {rmse:.4f}")
      results_regressors.append(["housing", name, rmse])

      # Classification Example
      X, y = load_iris(return_X_y=True)
      X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
      classifier = BoosterClassifier(obj=RegressorClass(), n_estimators=100, learning_rate=0.1,
                                    n_hidden_features=10, verbose=verbose, seed=42)
      start = time()
      try:
          classifier.fit(X_train, y_train)
      except Exception as e: # this is for Windows users
          y_train = y_train.astype('int32')
          classifier.fit(X_train, y_train)
      y_pred = classifier.predict(X_test)
      print(f"Elapsed: {time() - start} s")
      accuracy = accuracy_score(y_test, y_pred)
      print(f"Accuracy for classification: {accuracy:.4f}")
      results_classifiers.append(["iris", name, accuracy])

      X, y = load_wine(return_X_y=True)
      X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
      classifier = BoosterClassifier(obj=RegressorClass(), n_estimators=100, learning_rate=0.1,
                                    n_hidden_features=10, verbose=verbose, seed=42)
      start = time()
      try:
          classifier.fit(X_train, y_train)
      except Exception as e: # this is for Windows users
          y_train = y_train.astype('int32')
          classifier.fit(X_train, y_train)
      y_pred = classifier.predict(X_test)
      print(f"Elapsed: {time() - start} s")
      accuracy = accuracy_score(y_test, y_pred)
      print(f"Accuracy for classification: {accuracy:.4f}")
      results_classifiers.append(["wine", name, accuracy])

      X, y = load_breast_cancer(return_X_y=True)
      X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
      classifier = BoosterClassifier(obj=RegressorClass(), n_estimators=100, learning_rate=0.1,
                                    n_hidden_features=10, verbose=verbose, seed=42)
      start = time()
      try:
          classifier.fit(X_train, y_train)
      except Exception as e:
          y_train = y_train.astype('int32')
          classifier.fit(X_train, y_train)
      y_pred = classifier.predict(X_test)
      print(f"Elapsed: {time() - start} s")
      accuracy = accuracy_score(y_test, y_pred)
      print(f"Accuracy for classification: {accuracy:.4f}")
      results_classifiers.append(["breast_cancer", name, accuracy])

      X, y = load_digits(return_X_y=True)
      X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
      classifier = BoosterClassifier(obj=RegressorClass(), n_estimators=100, learning_rate=0.1,
                                    n_hidden_features=10, verbose=verbose, seed=42)
      start = time()
      try:
          classifier.fit(X_train, y_train)
      except Exception as e:
          y_train = y_train.astype('int32')
          classifier.fit(X_train, y_train)
      y_pred = classifier.predict(X_test)
      print(f"Elapsed: {time() - start} s")
      accuracy = accuracy_score(y_test, y_pred)
      print(f"Accuracy for classification: {accuracy:.4f}")
      results_classifiers.append(["digits", name, accuracy])
    except Exception as e:
      continue

  0%|          | 0/55 [00:00<?, ?it/s]

Running with ARDRegression
Elapsed: 4.067568063735962 s
RMSE for regression: 50.7652

  2%|▏         | 1/55 [00:05<04:50,  5.37s/it]

Elapsed: 0.5349180698394775 s
RMSE for regression: 0.4230

Running with AdaBoostRegressor
Elapsed: 13.95115351676941 s
RMSE for regression: 52.9127

  4%|▎         | 2/55 [00:40<20:10, 22.85s/it]

Elapsed: 21.075905323028564 s
RMSE for regression: 0.3281

Running with BaggingRegressor
Elapsed: 7.636256456375122 s
RMSE for regression: 54.6999
Elapsed: 19.565755367279053 s
RMSE for regression: 0.3489
Elapsed: 4.050839185714722 s
Accuracy for classification: 1.0000
Elapsed: 4.740557432174683 s
Accuracy for classification: 0.9722
Elapsed: 30.266514778137207 s
Accuracy for classification: 0.9561

  5%|▌         | 3/55 [02:56<1:04:47, 74.76s/it]

Elapsed: 70.22430944442749 s
Accuracy for classification: 0.9667

Running with BayesianRidge
Elapsed: 0.4146571159362793 s
RMSE for regression: 50.7598

  7%|▋         | 4/55 [02:57<38:45, 45.59s/it]

Elapsed: 0.43773674964904785 s
RMSE for regression: 0.4328

Running with CCA

Running with DecisionTreeRegressor
Elapsed: 0.7410914897918701 s
RMSE for regression: 61.0645
Elapsed: 1.557697057723999 s
RMSE for regression: 0.3849
Elapsed: 0.17411327362060547 s
Accuracy for classification: 1.0000
Elapsed: 0.23183941841125488 s
Accuracy for classification: 0.9444
Elapsed: 1.1447832584381104 s
Accuracy for classification: 0.9386

df_results_regressors = pd.DataFrame(results_regressors, columns=['Dataset', 'Model', 'RMSE'])
df_results_regressors.sort_values(by='RMSE', ascending=True, inplace=True)

df_results_classifiers = pd.DataFrame(results_classifiers, columns=['Dataset', 'Model', 'Accuracy'])
df_results_classifiers.sort_values(by='Accuracy', ascending=False, inplace=True)

df_results_regressors_diabetes = df_results_regressors[df_results_regressors['Dataset'] == 'diabetes']
df_results_regressors_housing = df_results_regressors[df_results_regressors['Dataset'] == 'housing']
df_results_classifiers_iris = df_results_classifiers[df_results_classifiers['Dataset'] == 'iris']
df_results_classifiers_wine = df_results_classifiers[df_results_classifiers['Dataset'] == 'wine']
df_results_classifiers_breast_cancer = df_results_classifiers[df_results_classifiers['Dataset'] == 'breast_cancer']
df_results_classifiers_digits = df_results_classifiers[df_results_classifiers['Dataset'] == 'digits']

print("Best regressors:")
display(df_results_regressors_diabetes)
display(df_results_regressors_housing)

print("\nBest classifiers:")
display(df_results_classifiers_iris)
display(df_results_classifiers_wine)
display(df_results_classifiers_breast_cancer)
display(df_results_classifiers_digits)

Best regressors:

Best classifiers:

!pip install nnetsauce

import nnetsauce as ns
import pandas as pd
from sklearn.utils import all_estimators
from tqdm import tqdm
from sklearn.utils.multiclass import type_of_target
from sklearn.datasets import fetch_california_housing


# Get all scikit-learn regressors
estimators = all_estimators(type_filter='regressor')

results_regressors = []
results_classifiers = []

verbose = 0

url = "https://raw.githubusercontent.com/Techtonique/"
url += "datasets/main/time_series/multivariate/"
url += "ice_cream_vs_heater.csv"
df_temp = pd.read_csv(url)
df_temp.index = pd.DatetimeIndex(df_temp.date)
# must have# first other difference
df_icecream = df_temp.drop(columns=['date']).diff().dropna()

for name, RegressorClass in tqdm(estimators):

    if name in ['AdaBoostRegressor', 'MultiOutputRegressor', 'MultiOutputClassifier', 'StackingRegressor', 'StackingClassifier',
                    'VotingRegressor', 'VotingClassifier', 'TransformedTargetRegressor', 'RegressorChain',
                    'GradientBoostingRegressor', 'HistGradientBoostingRegressor', 'RandomForestRegressor',
                    'ExtraTreesRegressor', 'MLPRegressor', 'TheilSenRegressor']:

        continue

    try:
      print(f"\nRunning with {name}")
      # Regression Examples
      mdl = BoosterRegressor(obj=RegressorClass(), n_estimators=100, learning_rate=0.1,
                                  n_hidden_features=10, verbose=verbose, seed=42)
      regr = ns.MTS(obj=mdl,
                    type_pi="scp2-kde",
                    replications=250,
                    lags=20,
                    show_progress=False)
      regr.fit(df_icecream)
      regr.predict(h=30)
      regr.plot("heater", type_plot="pi")
    except Exception as e:
      print(e)
      continue

  0%|          | 0/55 [00:00<?, ?it/s]

Running with ARDRegression

  2%|▏         | 1/55 [00:37<34:11, 37.98s/it]

Running with BaggingRegressor

  5%|▌         | 3/55 [01:16<20:55, 24.15s/it]

Running with BayesianRidge

  7%|▋         | 4/55 [01:21<14:54, 17.55s/it]

Running with CCA
`n_components` upper bound is 1. Got 2 instead. Reduce `n_components`.

Running with DecisionTreeRegressor

 11%|█         | 6/55 [01:26<08:14, 10.09s/it]

Running with DummyRegressor

Boosting any randomized based learner for regression, classification and univariate/multivariate time series forcasting

Boosting with Cybooster¶

1 - Simple example¶

2 - Loop on models¶

3 - Time series forecasting using `nnetsauce.MTS`¶

	Dataset	Model	RMSE
66	diabetes	TheilSenRegressor	49.002812
52	diabetes	PassiveAggressiveRegressor	49.017006
42	diabetes	LinearSVR	49.195826
64	diabetes	SVR	49.432314
62	diabetes	SGDRegressor	49.740770
50	diabetes	PLSRegression	49.877624
40	diabetes	LinearRegression	50.038024
24	diabetes	KernelRidge	50.137963
60	diabetes	RidgeCV	50.153038
58	diabetes	Ridge	50.180423
20	diabetes	HuberRegressor	50.354598
30	diabetes	Lasso	50.490869
34	diabetes	LassoLars	50.491092
68	diabetes	TweedieRegressor	50.589740
6	diabetes	BayesianRidge	50.759803
0	diabetes	ARDRegression	50.765231
12	diabetes	ElasticNet	50.773545
46	diabetes	OrthogonalMatchingPursuit	50.797879
48	diabetes	OrthogonalMatchingPursuitCV	50.799486
38	diabetes	LassoLarsIC	50.866918
14	diabetes	ElasticNetCV	51.243455
32	diabetes	LassoCV	51.333102
36	diabetes	LassoLarsCV	51.658117
44	diabetes	NuSVR	51.965976
2	diabetes	AdaBoostRegressor	52.912706
28	diabetes	LarsCV	52.936573
4	diabetes	BaggingRegressor	54.699888
16	diabetes	ExtraTreeRegressor	55.071533
56	diabetes	RANSACRegressor	58.883571
8	diabetes	DecisionTreeRegressor	61.064470
18	diabetes	GaussianProcessRegressor	66.618258
54	diabetes	QuantileRegressor	72.886227
10	diabetes	DummyRegressor	73.222493
22	diabetes	KNeighborsRegressor	219.818565
26	diabetes	Lars	249.197131

	Dataset	Model	RMSE
3	housing	AdaBoostRegressor	0.328082
65	housing	SVR	0.345998
17	housing	ExtraTreeRegressor	0.346823
45	housing	NuSVR	0.346886
5	housing	BaggingRegressor	0.348871
21	housing	HuberRegressor	0.373068
43	housing	LinearSVR	0.376118
67	housing	TheilSenRegressor	0.384006
9	housing	DecisionTreeRegressor	0.384891
41	housing	LinearRegression	0.396910
59	housing	Ridge	0.397799
61	housing	RidgeCV	0.402997
53	housing	PassiveAggressiveRegressor	0.412686
63	housing	SGDRegressor	0.421484
1	housing	ARDRegression	0.423014
39	housing	LassoLarsIC	0.423339
25	housing	KernelRidge	0.423757
51	housing	PLSRegression	0.426112
37	housing	LassoLarsCV	0.431140
7	housing	BayesianRidge	0.432840
15	housing	ElasticNetCV	0.434310
33	housing	LassoCV	0.434807
49	housing	OrthogonalMatchingPursuitCV	0.438809
69	housing	TweedieRegressor	0.444735
47	housing	OrthogonalMatchingPursuit	0.454569
29	housing	LarsCV	0.464425
57	housing	RANSACRegressor	0.472617
19	housing	GaussianProcessRegressor	0.501228
27	housing	Lars	0.548208
13	housing	ElasticNet	0.732468
55	housing	QuantileRegressor	0.852777
35	housing	LassoLars	0.869965
11	housing	DummyRegressor	0.869965
31	housing	Lasso	0.869965
23	housing	KNeighborsRegressor	1.000700

	Dataset	Model	Accuracy
0	iris	BaggingRegressor	1.000000
24	iris	KNeighborsRegressor	1.000000
4	iris	DecisionTreeRegressor	1.000000
52	iris	PLSRegression	1.000000
20	iris	GaussianProcessRegressor	1.000000
44	iris	LinearRegression	1.000000
64	iris	RidgeCV	1.000000
60	iris	Ridge	1.000000
48	iris	OrthogonalMatchingPursuit	0.966667
28	iris	KernelRidge	0.933333
16	iris	ExtraTreeRegressor	0.933333
56	iris	RANSACRegressor	0.500000
12	iris	ElasticNet	0.300000
8	iris	DummyRegressor	0.300000
40	iris	LassoLars	0.300000
36	iris	Lasso	0.300000
32	iris	Lars	0.266667

	Dataset	Model	Accuracy
17	wine	ExtraTreeRegressor	1.000000
53	wine	PLSRegression	1.000000
61	wine	Ridge	1.000000
29	wine	KernelRidge	1.000000
49	wine	OrthogonalMatchingPursuit	1.000000
65	wine	RidgeCV	1.000000
45	wine	LinearRegression	1.000000
1	wine	BaggingRegressor	0.972222
25	wine	KNeighborsRegressor	0.972222
5	wine	DecisionTreeRegressor	0.944444
21	wine	GaussianProcessRegressor	0.833333
33	wine	Lars	0.777778
13	wine	ElasticNet	0.500000
37	wine	Lasso	0.388889
41	wine	LassoLars	0.388889
57	wine	RANSACRegressor	0.388889
9	wine	DummyRegressor	0.388889

	Dataset	Model	Accuracy
50	breast_cancer	OrthogonalMatchingPursuit	0.982456
54	breast_cancer	PLSRegression	0.973684
30	breast_cancer	KernelRidge	0.964912
66	breast_cancer	RidgeCV	0.964912
46	breast_cancer	LinearRegression	0.964912
62	breast_cancer	Ridge	0.964912
18	breast_cancer	ExtraTreeRegressor	0.964912
2	breast_cancer	BaggingRegressor	0.956140
6	breast_cancer	DecisionTreeRegressor	0.938596
26	breast_cancer	KNeighborsRegressor	0.885965
58	breast_cancer	RANSACRegressor	0.842105
14	breast_cancer	ElasticNet	0.824561
38	breast_cancer	Lasso	0.815789
42	breast_cancer	LassoLars	0.815789
22	breast_cancer	GaussianProcessRegressor	0.631579
10	breast_cancer	DummyRegressor	0.622807
34	breast_cancer	Lars	0.526316

	Dataset	Model	Accuracy
19	digits	ExtraTreeRegressor	0.975000
3	digits	BaggingRegressor	0.966667
51	digits	OrthogonalMatchingPursuit	0.958333
27	digits	KNeighborsRegressor	0.958333
31	digits	KernelRidge	0.952778
47	digits	LinearRegression	0.950000
63	digits	Ridge	0.950000
67	digits	RidgeCV	0.950000
55	digits	PLSRegression	0.930556
7	digits	DecisionTreeRegressor	0.858333
59	digits	RANSACRegressor	0.530556
35	digits	Lars	0.300000
23	digits	GaussianProcessRegressor	0.227778
11	digits	DummyRegressor	0.077778
15	digits	ElasticNet	0.077778
39	digits	Lasso	0.077778
43	digits	LassoLars	0.077778

Boosting any randomized based learner for regression, classification and univariate/multivariate time series forcasting

Boosting with Cybooster¶

1 - Simple example¶

2 - Loop on models¶

3 - Time series forecasting using nnetsauce.MTS¶

3 - Time series forecasting using `nnetsauce.MTS`¶