机器学习的数据处理与逻辑回归预测股市

import numpy as np
import pandas as pd


def produce_data(n_points=1000):
    # 创建储存数据的DataFrame
    data = pd.DataFrame()
    
    # 生成虚拟研究数据
    np.random.seed(0)
    # 生成学习特征
    data['practice'] = np.random.random(n_points)
    data['time_period'] = np.random.random(n_points)
    data['error'] = np.random.random(n_points)
    # 生成类别标签
    data['label'] = np.round(data['practice']*data['time_period'] + 0.3 + 0.1*data['error'])
    data['label'][data['practice']>0.8] = 1
    data['label'][data['time_period']>0.8] = 1
    
    return(data)
    
if __name__ == '__main__':
    data = produce_data()

from sklearn.model_selection import train_test_split
features_train,features_test,label_train,label_test = 
train_test_split(data[['practice','time_period']],data['label'],test_size=0.25 )             #训练集比例

train_data = features_train.copy()             #防止篡改
train_data['label'] = label_train
train_data.head()


practice_fail = train_data[train_data['label']==0]['practice']
time_preiod_fail = train_data[train_data['label']==0]['time_period']
# practice_fail.head()
# time_preiod_fail.head()
practice_pass = train_data[train_data['label']==1]['practice']
time_preiod_pass = train_data[train_data['label']==1]['time_period']
plt.figure()           #建一个画图窗口
plt.xlim(0.0,1.0)            #x轴
plt.ylim(0.0,1.0)
plt.scatter(practice_fail,time_preiod_fail,color = 'b',label = 'fail') #散点图
plt.scatter(practice_pass,time_preiod_pass,color = 'r',label = 'pass')
plt.xlabel('practice')
plt.ylabel('time_period')
plt.legend()      

from sklearn.linear_model import LogisticRegression


clf_LR = LogisticRegression()                   #定义逻辑回归的分类器    初始化模型
clf_LR.fit(features_train,label_train)           #重点  #用训练集进行训练
pred_LR = clf_LR.predict(features_test)               # 预测 测试集   返回预测label


plot_pic(clf_LR,features_test,label_test)             # clf_LR 训练好模型为参数为传入 画图

###plot_pic函数
import numpy as np
import matplotlib.pyplot as plt




def plot_pic(clf, X_test, y_test):
    x_min = 0.0; x_max = 1.0
    y_min = 0.0; y_max = 1.0
    
    # 画出决策边界，我们为每一个点绘制一个颜色


    step = .01  # 设置步长
    xx, yy = np.meshgrid(np.arange(x_min, x_max, step), np.arange(y_min, y_max, step))
    Z = clf.predict(np.c_[xx.ravel(), yy.ravel()])


    # 绘图
    Z = Z.reshape(xx.shape)
    plt.xlim(xx.min(), xx.max())
    plt.ylim(yy.min(), yy.max())


    plt.pcolormesh(xx, yy, Z, cmap=plt.cm.RdYlBu)




    data = X_test.copy()
    data['label'] = y_test
    
    practice_fail = data[data['label']==0]['practice']
    time_period_fail = data[data['label']==0]['time_period']
    practice_pass = data[data['label']==1]['practice']
    time_period_pass = data[data['label']==1]['time_period']




    plt.scatter(practice_fail, time_period_fail, color = 'b', label='fail')
    plt.scatter(practice_pass, time_period_pass, color = 'r', label='pass')
    plt.legend()
    plt.xlabel('practice')
    plt.ylabel('time_period')
    plt.legend(loc='upper right')

from sklearn.metrics import accuracy_score
acc = accuracy_score(pred_LR,label_test)
acc

from sklearn.tree import DecisionTreeClassifier


clf_DT = DecisionTreeClassifier()
clf_DT.fit(features_train,label_train)
pred_DT = clf_DT.predict(features_test)               # 预测 测试集   返回预测label


plot_pic(clf_DT,features_test,label_test)             # clf_LR 训练好模型为参数为传入 画图
from sklearn.metrics import accuracy_score
acc = accuracy_score(pred_DT,label_test)
acc

from sklearn.ensemble import RandomForestClassifier


clf_RF = RandomForestClassifier()
clf_RF.fit(features_train,label_train)
pred_RF = clf_RF.predict(features_test)               # 预测 测试集   返回预测label


plot_pic(clf_RF,features_test,label_test)             # clf_LR 训练好模型为参数为传入 画图


from sklearn.metrics import accuracy_score
acc = accuracy_score(pred_RF,label_test)
acc

#KNN算法原理
from sklearn.neighbors import KNeighborsClassifier


clf_KNN = KNeighborsClassifier(n_neighbors=7)
clf_KNN.fit(features_train,label_train)
pred_KNN = clf_KNN.predict(features_test)               # 预测 测试集   返回预测label


plot_pic(clf_KNN,features_test,label_test)             # clf_LR 训练好模型为参数为传入 画图


from sklearn.metrics import accuracy_score
acc = accuracy_score(pred_KNN,label_test)
acc

from sklearn.svm import SVC


clf_SVC = SVC(kernel='rbf',gamma = 100)          #默认kernel   = rbf  高斯核   linear 线性核   ploy  多项式核 gamma SVM超参数（根据经验）
clf_SVC.fit(features_train,label_train)           #gamma 控制SVM中点距离大小的影响程度
pred_SVC = clf_SVC.predict(features_test)               # 预测 测试集   返回预测label


plot_pic(clf_SVC,features_test,label_test)             # 训练好模型为参数为传入 画图


from sklearn.metrics import accuracy_score
acc = accuracy_score(pred_SVC,label_test)
acc