Examples for Evolutionary Automated Machine Learning¶

Evolutionary Bagging Classifier¶

This classifier takes the best performing one for the mutation step

In [1]:

from river import metrics, evaluate
from river.datasets import Phishing
from EvOAutoML.classification import EvolutionaryBaggingClassifier
from tqdm import tqdm

from river import metrics, evaluate from river.datasets import Phishing from EvOAutoML.classification import EvolutionaryBaggingClassifier from tqdm import tqdm

In [2]:

n_samples = 10000
dataset = Phishing()

metric = metrics.Accuracy()
model = EvolutionaryBaggingClassifier()

n_samples = 10000 dataset = Phishing() metric = metrics.Accuracy() model = EvolutionaryBaggingClassifier()

In [3]:

for x,y in tqdm(dataset):
    y_pred = model.predict_one(x)      # make a prediction
    metric = metric.update(y, y_pred)  # update the metric
    model = model.learn_one(x, y)    # make the model learn
print(f'Accuracy: {metric.get()}')

for x,y in tqdm(dataset): y_pred = model.predict_one(x) # make a prediction metric = metric.update(y, y_pred) # update the metric model = model.learn_one(x, y) # make the model learn print(f'Accuracy: {metric.get()}')

1250it [00:13, 92.91it/s]

Accuracy: 0.8928

Evolutionary Oldest Bagging Classifier¶

This classifier removes the oldest ML pipeline from the ensemble.

In [4]:

from river import metrics, evaluate
from river.datasets import Phishing
from EvOAutoML.classification import EvolutionaryOldestBaggingClassifier
from tqdm import tqdm

from river import metrics, evaluate from river.datasets import Phishing from EvOAutoML.classification import EvolutionaryOldestBaggingClassifier from tqdm import tqdm

In [5]:

n_samples = 10000
dataset = Phishing()

metric = metrics.Accuracy()
model = EvolutionaryOldestBaggingClassifier()

n_samples = 10000 dataset = Phishing() metric = metrics.Accuracy() model = EvolutionaryOldestBaggingClassifier()

In [6]:

for x, y in tqdm(dataset):
    y_pred = model.predict_one(x)  # make a prediction
    metric = metric.update(y, y_pred)  # update the metric
    model = model.learn_one(x, y)  # make the model learn
print(f'Accuracy: {metric.get()}')

for x, y in tqdm(dataset): y_pred = model.predict_one(x) # make a prediction metric = metric.update(y, y_pred) # update the metric model = model.learn_one(x, y) # make the model learn print(f'Accuracy: {metric.get()}')

1250it [00:13, 91.51it/s]

Accuracy: 0.8944