分类准确度、超参数、网格搜索、归一化等

准确度：计算Y_True / Y_predict之间的准确度

import numpy as np
from matplotlib import pyplot as plt
from sklearn import datasets
digits= datasets.load_digits()
X = digits.data
from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
knn = KNeighborsClassifier(n_neighbors=3)
knn.fit(X_train, y_train)
y_predict = knn.predict(X_test)
# 准确度
sum(y_test == y_predict) / y_test.shape[0]
# sklearn库的准确度
knn.score(X_test, y_test)
# sklearn准确度的函数
from sklearn.metrics import accuracy_score
accuracy_score(y_test, y_predict)

超参数

什么是超参数：在运行机器学习算法之前需要设置的参数。

寻找最好的K

best_k = 1
best_score =0.0
weights = None
for k in range(1, 11):
    for weight in ['uniform', 'distance']:
        knn1 = KNeighborsClassifier(n_neighbors=k, weights=weight)
        knn1.fit(X_train, y_train)
        score = knn1.score(X_test, y_test)
        if score > best_score:
            weights = weight
            best_score = score
            best_k = k
print('best_k:%s' % best_k)
print('best_score:%s' % best_score)
print('best_weights:%s' % weights)

K近邻算法考虑距离的权重— 取距离的倒数。

网格搜索¶

param_grid = [
    {
        'weights': ['uniform'],
        'n_neighbors': [i for i in range(1, 11)]
    },
    {
        'weights': ['distance'],
        'n_neighbors': [i for i in range(1, 11)],
        'p': [i for i in range(1, 5)],
    },
]
from sklearn.model_selection import GridSearchCV
grid = GridSearchCV(KNeighborsClassifier(), param_grid, n_jobs=4)
grid.fit(X_train, y_train)
# 最佳模型
grid.best_estimator_

最值归一化

什么是归一化：将数据集映射到0-1之间 为什么：量纲不同导致距离的计算会偏向某一个个特征。 适用于有明显的边界。例如分数。受outlier影响较大

x = np.random.randint(0, 100, size=100)
(x - np.min(x)) / (np.max(x) - np.min(x))

均值方差归一化

把所有的数据归一到均值是0方差为1的分布中

std_data = (x - np.mean(x)) / np.std(x)

机器学习库的均值方差归一化和最值归一化

from sklearn.preprocessing import StandardScaler, MinMaxScaler
std = StandardScaler()
std.fit_transform(x.reshape(-1, 1))