lemon 🍋🍋🍋

一个轻量级的机器学习框架（纯python+numpy实现的迷你版scikit-learn）

examples

from lemon.datasets import load_iris
from lemon.model_utils.model_selection import train_test_split
from lemon.supervised.naive_bayes import GaussianNB
from lemon.model_utils.metrics import accuracy


x, y = load_iris(x_y=True)
x_train, x_test, y_train, y_test = train_test_split(x, y, split_rate=0.8, random_state=2020)

model = GaussianNB()
model.fit(x_train, y_train)
pred = model.predict(x_test)

print(accuracy(y_test, pred))

当前进度

• datasets(boston, breast_canner, iris, titanic, wine)
• processing
  • preprocessing(binarizer, transformer, discretizer, encoding, scaler)
  • decomposition(pca, svd)
  • impute(simple-impute)
• feature_utils
  • feature_selection (woe_iv)
• model_utils
  • metrics
  • model_selection(psi)
• supervised
  • linear_model(simple-linear, lasso, ridge, elastic-net, perceptron, logistic-regression)
  • naive_bayes(gaussian, multinomial)
  • neighbors(kd-tree-based-knn)
  • svm
  • tree
  • hmm
  • crf
  • ensemble
• semi_supervised
  • label_propagation
  • louvain
  • pagerank
• unsupervised
  • dbscan
  • kmeans(simple-kmeans, kmeans++, mini-batch-kmeans)