from sklearn import ensemble, datasets
import numpy as np
import matplotlib
from matplotlib import pyplot as plt

from torchvision.datasets import EMNIST

train = EMNIST(root="/Users/Noah QH/Downloads/emnist/", train=True, split="digits", download=True)
test = EMNIST(root="/Users/Noah QH/Downloads/emnist/", train=False, split="digits", download=True)

x_train, y_train = train.data.flatten(1, 2), train.targets
x_test, y_test = test.data.flatten(1, 2), test.targets

plt.figure(figsize=(12, 1))
plt.title("Input data examples", pad=30)
plt.axis('off')
colormap = matplotlib.colormaps["Greys"]
for i in range(10):
  plt.subplot(1, 10, i + 1)
  plt.box(True)
  plt.xticks([])
  plt.yticks([])
  plt.grid(True)
  plt.title(f'{y_train[i]}')
  plt.imshow(np.resize(x_train[i], (28, 28)).T, cmap=colormap)
plt.show()

rfc = ensemble.RandomForestClassifier(n_estimators=15, max_depth=None)
bagged_rfc = ensemble.BaggingRegressor(rfc, n_jobs=-1)
bagged_rfc.fit(x_train, y_train)

BaggingRegressor(estimator=RandomForestClassifier(n_estimators=15), n_jobs=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

y_pred = bagged_rfc.predict(x_test).astype(int).flatten()

error = (np.array(y_test) == y_pred)
correct = sum(error)
incorrect = len(y_test) - correct
accuracy = correct / (correct + incorrect)

plt.figure(figsize=(10, 14))
plt.title("Predictions", pad=20)
plt.axis('off')
colormap = matplotlib.colormaps["Greys"]
for i in range(100):
  plt.subplot(10, 10, i + 1)
  plt.box(True)
  plt.xticks([])
  plt.yticks([])
  plt.grid(True)
  plt.title(f'{y_pred[i]} | {y_test[i]}')
  plt.imshow(np.resize(x_test[i], (28, 28)).T, cmap=colormap)
plt.show()

print(f'Accuracy: {accuracy} ({correct} / {correct + incorrect} correct)')

Accuracy: 0.947575 (37903 / 40000 correct)