模型能直接应用吗？

得到的模型靠谱吗，能否和真实环境的数据拟合的很好？
image.png

解决方案

数据集切分
image.png

数据准备

image.png
image.png

切分数据集

数据集切分成两部分：训练集和测试集
image.png

封装切分算法

  1. # playML/model_selection.py
  2. import numpy as np
  5. def train_test_split(X, y, test_ratio=0.2, seed=None):
  6.     """将数据 X 和 y 按照test_ratio分割成X_train, X_test, y_train, y_test"""
  7.     assert X.shape[0] == y.shape[0], \
  8.         "the size of X must be equal to the size of y"
  9.     assert 0.0 <= test_ratio <= 1.0, \
  10.         "test_ration must be valid"
  12.     if seed:
  13.         np.random.seed(seed)
  15.     shuffled_indexes = np.random.permutation(len(X))
  17.     test_size = int(len(X) * test_ratio)
  18.     test_indexes = shuffled_indexes[:test_size]
  19.     train_indexes = shuffled_indexes[test_size:]
  21.     X_train = X[train_indexes]
  22.     y_train = y[train_indexes]
  24.     X_test = X[test_indexes]
  25.     y_test = y[test_indexes]
  27.     return X_train, X_test, y_train, y_test

image.png

模型准确度

image.png

sklearn中的切分算法

image.png

  1. # 关键源码
  2. from sklearn.model_selection import train_test_split
  4. X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=666)