machine_learn/linear/train_jqxx_index.py

#!/usr/bin/python
# -*- coding: UTF-8 -*-
'''
最简单的mse
'''
import sys
import os
sys.path.append(os.path.abspath('..'))
from util.config import config
import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import GridSearchCV
from sklearn.svm import SVC
from sklearn import metrics
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from imblearn.over_sampling import RandomOverSampler
import joblib


def curce_data(x,y,y_pred):
    x=x.tolist()
    y=y.tolist()
    y_pred=y_pred.tolist()
    results=zip(x,y,y_pred)
    results=["{},{},{}".format(s[0],s[1][0],s[2][0]) for s in results ]
    return results


def read_data(path):
    with open(path) as f :
        lines=f.readlines()
    lines=[eval(line.strip()) for line in lines]
    X,z,y=zip(*lines)
    X=np.array(X)
    y=np.array(y)
    return X,y


def demo(file, model_file):
    X_train,y_train=read_data(file)
    Xtrain, Xtest, Ytrain, Ytest = train_test_split(X_train, y_train, test_size=0.3)
    # 一个对象，它代表的线性回归模型，它的成员变量，就已经有了w，b. 刚生成w和b的时候 是随机的
    ros = RandomOverSampler(random_state=22)
    X_rsampled, y_resampled = ros.fit_resample(Xtrain, Ytrain)
    model = LogisticRegression(max_iter=1200)
    # 一调用这个函数，就会不停地找合适的w和b 直到误差最小
    model.fit(X_rsampled, y_resampled)
    # 打印W
    # print(model.coef_)
    # 打印b
    print(model.intercept_)
    # 模型已经训练完毕,用模型看下在训练集的表现
    # y_pred_train = model.predict(Xtrain)
    # # sklearn 求解训练集的mse
    # # y_train 在训练集上 真实的y值
    # # y_pred_train 通过模型预测出来的y值
    # # 计算  (y_train-y_pred_train)^2/n
    # train_mse = metrics.mean_squared_error(Ytrain, y_pred_train)
    # print("train准确率:", accuracy_score(y_pred_train, Ytrain))
    #
    # # 看下在测试集上的效果
    y_pred_test = model.predict(Xtest)
    print("test准确率:", accuracy_score(y_pred_test, Ytest))
    # 保存模型
    joblib.dump(model,  model_file)


def demo_1(file, model_file):
    X_train,y_train=read_data(file)
    Xtrain, Xtest, Ytrain, Ytest = train_test_split(X_train, y_train, test_size=0.25)
    # 随机过采样
    ros = RandomOverSampler(random_state=22)
    X_rsampled, y_resampled = ros.fit_resample(X_train, y_train)
    model = SVC(kernel='linear')
    # 一调用这个函数，就会不停地找合适的w和b 直到误差最小
    model.fit(X_rsampled, y_resampled)
    # 打印W
    # print(model.coef_)
    # 打印b
    print(model.intercept_)
    # # 模型已经训练完毕,用模型看下在训练集的表现
    # y_pred_train = model.predict(Xtrain)
    # # sklearn 求解训练集的mse
    # # y_train 在训练集上 真实的y值
    # # y_pred_train 通过模型预测出来的y值
    # # 计算  (y_train-y_pred_train)^2/n
    # train_mse = metrics.mean_squared_error(Ytrain, y_pred_train)
    # print("train准确率:", accuracy_score(y_pred_train, Ytrain))
    #
    # # 看下在测试集上的效果
    y_pred_test = model.predict(Xtest)
    print("test准确率:", accuracy_score(y_pred_test, Ytest))
    # # 保存模型
    joblib.dump(model,  model_file)


from sklearn.ensemble import RandomForestClassifier
def demo_2(file, model_file):
    X_train,y_train=read_data(file)
    Xtrain, Xtest, Ytrain, Ytest = train_test_split(X_train, y_train, test_size=0.25)
    rfc = RandomForestClassifier(random_state=0,n_estimators=20,max_depth=4)
    rfc = rfc.fit(Xtrain,Ytrain)
    print(rfc.score(Xtest,Ytest))
    # 看下在测试集上的效果
    y_pred_test = rfc.predict(Xtest)
    print("test准确率:", accuracy_score(y_pred_test, Ytest))
    # 保存模型
    joblib.dump(rfc,  model_file)


from sklearn.ensemble import GradientBoostingClassifier
def demo_4(file, model_file):
    X_train,y_train=read_data(file)
    # Xtrain, Xtest, Ytrain, Ytest = train_test_split(X_train, y_train, test_size=0.3)
    param_test1 = {'max_depth':range(3,13, 2), "min_samples_split":range(80, 800, 50)}
    gbm2 = GradientBoostingClassifier(n_estimators=130, learning_rate=0.7,
                                     max_features='sqrt', random_state=10)
    gseach1 = GridSearchCV(estimator=gbm2, param_grid=param_test1, scoring="roc_auc", cv = 5)
    gseach1.fit(X_train,y_train)
    print(gseach1.param_grid, gseach1.best_params_,  gseach1.best_score_)
    # rfc = gbm2.fit(Xtrain,Ytrain)
    # print(gbm2.score(Xtest,Ytest))
    # # 看下在测试集上的效果
    # y_pred_test = gbm2.predict(Xtest)
    # print("test准确率:", accuracy_score(y_pred_test, Ytest))
    # # 保存模型
    # joblib.dump(gbm2,  model_file)

def demo_3(file, model_file):
    X_train,y_train=read_data(file)
    # Xtrain, Xtest, Ytrain, Ytest = train_test_split(X_train, y_train, test_size=0.3)
    # gbm2 = GradientBoostingClassifier(n_estimators=120, max_depth=4, min_samples_split=180, learning_rate=0.7,
    #                                   max_features='sqrt', random_state=101)
    gbm2 = GradientBoostingClassifier(n_estimators=15, max_depth=4, min_samples_split=140, learning_rate=0.7,
                                      max_features='sqrt', random_state=101)
    # gbm2 = GradientBoostingClassifier(n_estimators=30, max_depth=4, min_samples_split=200, learning_rate=0.7,
    #                                   max_features='sqrt', random_state=101)
    rfc = gbm2.fit(X_train,y_train)
    print(gbm2.feature_importances_)
    # print(gbm2.score(Xtest,Ytest))
    # 看下在测试集上的效果
    # y_pred_test = gbm2.predict(Xtest)
    # print("test准确率:", accuracy_score(y_pred_test, Ytest))
    # 保存模型
    joblib.dump(gbm2,  model_file)


if __name__ == '__main__':
    root_dir = 'D:\\data\\quantization\\jqxx2\\'
    model_dir = 'D:\\data\\quantization\\jqxx2_tree_model\\'
    m = '399325.SZ.log' #12
    demo_3(root_dir + m, model_dir + str(m)[:6] + '.pkl')