当前位置：首页 > news >正文

机器学习算法-朴素贝叶斯（附带拉普拉斯平滑处理）

news 2025/7/30 7:04:28

朴素贝叶斯-python实现

数据集 excel文件

在这里插入图片描述

具体实现代码如下：

import pandas as pd
import numpy as npclass NavieBayes:def __init__(self):self.train_x = Noneself.train_y = Nonedef fit(self, X, Y, Labels):self.train_x = Xself.train_y = Ydef predict(self, x):uni_y = np.unique(self.train_y)p, n = 0.0, ""for res in uni_y:size = self.train_y.count(res)p_fz, p_fm = size / len(self.train_y), 1.0for xindex, xvalue in enumerate(x):c = [vx[xindex] for vx, vy in zip(self.train_x, self.train_y) if vy == res and vx[xindex] == xvalue]p_fz *= (len(c) / size)p_fm *= (len([item for item in self.train_x if item[xindex] == xvalue]) / len(self.train_y))if p_fz / p_fm > p:p = p_fz / p_fmn = resreturn n# 划分数据集
def split_test_data(path):df = pd.read_excel(path)x, y, labels = [], [], []labels = df.columns.tolist()res = labels.pop()y = df[res].tolist()for item in range(df.shape[0]):x.append(df.loc[item].tolist()[:len(labels)])return x, y, labelsX, Y, Labels = split_test_data("your excel file path")bayes_model = NavieBayes()
bayes_model.fit(X, Y, Labels)
# 输入 长相、性格、身高、上进 4个属性进行预测
print(f"贝叶斯预测:{bayes_model.predict(["帅", "好", "高","上进"])}")

引入拉普拉斯平滑

在这里插入图片描述

具体实现代码如下：

import numpy as np
import pandas as pddef split_test_data(path):df = pd.read_excel("./test_data.xlsx")x, y, labels = [], [], []labels = df.columns.tolist()res = labels.pop()y = df[res].tolist()for item in range(df.shape[0]):x.append(df.loc[item].tolist()[:len(labels)])return x, y, labelsX, Y, Labels = split_test_data("df")class NavieBayes:def __init__(self):self.train_x = []self.train_y = []self.labels = []def fit(self, x, y, l):self.train_x = xself.train_y = yself.labels = ldef predict(self, x):uni_y = np.unique(self.train_y)p, n = 0.0, " "for res in uni_y:size = self.train_y.count(res)p_fz, p_fm = size / len(self.train_y), 1.0for xindex, xvalue in enumerate(x):c = [vx[xindex] for vx, vy in zip(self.train_x, self.train_y) if vy == res and vx[xindex] == xvalue]d = [item[xindex] for item in self.train_x]+[x[xindex]]p_fz *= (len(c) / size+len(np.unique(d)))if p_fz > p:p = p_fzn = resreturn nbayes_model = NavieBayes()
bayes_model.fit(X, Y, Labels)
print(bayes_model.predict(['帅','巨好','特高','不上进']))