分别基于贝叶斯、BP神经网络、KNN进行邮件入侵分类检测python源码+详细项目说明+数据集+模型.zip

import pandas as pd import numpy as np import torch from time import time, strftime, localtime, clock from sklearn import metrics from sklearn.model_selection import train_test_split from sklearn.neighbors import KNeighborsClassifier from sklearn.naive_bayes import GaussianNB from sklearn.tree import DecisionTreeClassifier from sklearn.metrics import roc_curve, auc from sklearn.metrics import precision_recall_curve from sklearn.metrics import confusion_matrix from sklearn.metrics import accuracy_score from sklearn.metrics import precision_score from sklearn.metrics import recall_score from sklearn.metrics import f1_score import joblib from classify_knn import Knn from bayes import Gaussian_Bayes from BP import Net from tree import Tree from pyecharts.charts import Line from pyecharts.charts import Bar import pyecharts.options as opts import matplotlib.pyplot as plt class Classify(): def __init__(self, category=5): self.datafile = 'data/kddcup.data_10_percent_corrected_save_8w.csv' # self.datafile = 'data/corrected_save2.csv' self.category = category plt.rcParams['font.family'] = ['sans-serif'] plt.rcParams['font.sans-serif'] = ['SimHei'] plt.rcParams['axes.unicode_minus']=False self.time=strftime("%Y-%m-%d_%H-%M-%S", localtime()) def train_test(self): df_all = pd.read_csv(self.datafile) drop_name = [str(i) for i in range(31, 42)] y = np.array(df_all['41']) # 表示分类的列 df_all = df_all.drop(drop_name, axis=1) # 删除后十个 x = np.array(df_all) self.train_data, self.test_data, self.train_target, self.test_target = train_test_split(x, y, test_size=0.2, random_state=1) # 划分数据集 # 处理为二分类 def classify_two(self): for i in range(len(self.train_target)): if self.train_target[i]!=0: self.train_target[i]=1 # 不是正常的标签就置为1 for i in range(len(self.test_target)): if self.test_target[i]!=0: self.test_target[i]=1 # 不是正常的标签就置为1 def classify(self): model_evaluation = {} model_confusion = {} model_score = {} path = 'model/'+str(self.category)+'_' print("真实结果：",self.test_target) # knn = Knn() # k = knn.choose_K(5, self.train_data, self.train_target) # print("最终选择的k为："+str(k)) # knn = Knn(3) # knn.fit(self.train_data, self.train_target) # knn_result, knn_result_pro = knn.predict(self.test_data) # print("KNN预测结果：",knn_result) # knn_evaluation = self.Evaluation(self.test_target, knn_result) # model_evaluation['KNN']=knn_evaluation # knn = KNeighborsClassifier(n_neighbors=3) # knn.fit(self.train_data, self.train_target) # joblib.dump(knn, path+"knn_8w.pkl") knn = joblib.load(path+"knn_8w.pkl") knn_result = knn.predict(self.test_data) knn_result_pro = knn.predict_proba(self.test_data) print("KNN预测结果：",knn_result) knn_evaluation, knn_confu = self.Evaluation(self.test_target, knn_result) model_evaluation['KNN']=knn_evaluation model_confusion['KNN']=knn_confu model_score['KNN']=knn_result_pro # bayes = GaussianNB() # bayes.fit(self.train_data, self.train_target) # joblib.dump(bayes, path+"bayes_8w.pkl") bayes = joblib.load(path+"bayes_8w.pkl") bayes_result = bayes.predict(self.test_data) bayes_result_pro = bayes.predict_proba(self.test_data) print("贝叶斯预测结果：",bayes_result) bayes_evaluation, bayes_confu = self.Evaluation(self.test_target, bayes_result) model_evaluation['贝叶斯']=bayes_evaluation model_confusion['贝叶斯']=bayes_confu model_score['贝叶斯']=bayes_result_pro # bp = Net(self.train_data.shape[1], [20], self.category) # bp.train(self.train_data, self.train_target, 10000) # torch.save(bp,path+'bp_8w.pth') bp = torch.load(path+'bp_8w.pth') bp_result, bp_result_pro = bp.test(self.test_data) print("BP神经网络预测结果：",bp_result) bp_evaluation, bp_confu = self.Evaluation(self.test_target, bp_result) model_evaluation['BP神经网络']=bp_evaluation model_confusion['BP神经网络']=bp_confu model_score['BP神经网络']=bp_result_pro # tree = Tree() # tree.choose(self.train_data, self.train_target, self.test_data, self.test_target) # tree = DecisionTreeClassifier() # tree.fit(self.train_data, self.train_target) # joblib.dump(tree, path+"tree_8w.pkl") tree = joblib.load(path+"tree_8w.pkl") tree_result = tree.predict(self.test_data) tree_result_pro = tree.predict_proba(self.test_data) print("决策树预测结果：",tree_result) tree_result = tree_result.astype(int) tree_evaluation, tree_confu = self.Evaluation(self.test_target, tree_result) model_evaluation['决策树']=tree_evaluation model_confusion['决策树']=tree_confu model_score['决策树']=tree_result_pro self.draw_evaluation(model_evaluation) self.draw_confusion(model_confusion) multi = True if self.category==2: multi = False self.ROC(self.test_target, model_score, multi=multi) self.P_R(self.test_target, model_score, multi=multi) def Evaluation(self, target, test): self.loss = [] for i in range(self.category): if i not in target and i not in test: self.loss.append(i) result = [] # 混淆矩阵 confusion = confusion_matrix(target, test) if len(self.loss)!=0: for l in self.loss: b = np.zeros(confusion.shape[0]) confusion = np.insert(confusion, l, b, axis=1) b = np.zeros(confusion.shape[1]) confusion = np.insert(confusion, l, b, axis=0) # 准确率 ACC = round(100*(accuracy_score(target, test)),2) result.append(ACC) # 精确率 P = round(100*(precision_score(target, test, average="weighted")),2) result.append(P) # 召回率 R = round(100*(recall_score(target, test, average="weighted")),2) result.append(R) # F1-Score F = round(100*(f1_score(target, test, average="weighted")),2) result.append(F) return result, confusion def ROC(self, target, model_score, multi=False): fig = plt.figure(figsize=(8,8)) for model, socre in model_score.items(): if multi: new_target = [] for t in target: y = np.zeros(self.category-len(self.loss)) y[t]=1 new_target.append(y) new_target = np.array(new_target) fpr = {} tpr = {} roc_auc = {} for i in range(self.category-len(self.loss)): fpr[i], tpr[i], _ = roc_curve(new_target[:, i], socre[:, i]) roc_auc[i] = auc(fpr[i], tpr[i]) all_fpr = np.unique(np.concatenate([fpr[i] for i in range(self.category-len(self.loss))])) mean_tpr = np.zeros_like(all_fpr) for i in range(self.category-len(self.loss)): mean_tpr += np.interp(all_fpr, fpr[i], tpr[i]) mean_tpr /= (self.category-len(self.loss)) FPR = all_fpr TPR = mean_tpr else: