基于回归模型的销售预测

# 导入库
import pandas as pd
import numpy as np
from sklearn.linear_model import BayesianRidge, ElasticNet 
from sklearn.svm import SVR 
from xgboost import XGBRegressor
from sklearn.ensemble import GradientBoostingRegressor 
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import explained_variance_score, mean_absolute_error, \
mean_squared_error, r2_score 
import matplotlib.pyplot as plt 
from sklearn.model_selection import GridSearchCV 

# 读取数据
raw_data = pd.read_csv('regression.txt', delimiter=' ', header=None)  # 读取数据文件
raw_data.head()

# 拆分因变量
X_raw,y = raw_data.iloc[:, :-1],raw_data.iloc[:, -1]  # 分割自变量,因变量

# 数据标准化
model_ss = StandardScaler()
X = model_ss.fit_transform(X_raw)
X = pd.DataFrame(X, columns=raw_data.columns[:-1])

# 样本拆分
num = int(X.shape[0]*0.7)
X_train,X_test = X.iloc[:num,:],X.iloc[num:,:] # 拆分训练集和测试集
y_train,y_test = y[:num],y[num:] # 拆分训练集和测试集

# 初选回归模型
model_names = ['BayesianRidge', 'XGBR', 'ElasticNet', 'SVR', 'GBR']  # 不同模型的名称列表
model_br = BayesianRidge()  # 贝叶斯岭回归
model_xgbr = XGBRegressor(random_state=0)  # XGBR
model_etc = ElasticNet(random_state=0)  # 弹性网络回归
model_svr = SVR(gamma='scale')  # 支持向量机回归
model_gbr = GradientBoostingRegressor(random_state=0)  # 梯度增强回归
model_list = [model_br, model_xgbr, model_etc,model_svr, model_gbr]
pre_y_list = [model.fit(X_train, y_train).predict(X_test) for model in model_list]  # 各个回归模型预测的y值列表

# 模型效果评估
n_samples, n_features = X.shape  # 总样本量,总特征数
model_metrics_functions = [explained_variance_score, mean_absolute_error, mean_squared_error,r2_score]  # 回归评估指标对象集
model_metrics_list = [[m(y_test, pre_y_list[i]) for m in model_metrics_functions] for i in range(len(model_list))]  # 回归评估指标列表
regresstion_score = pd.DataFrame(model_metrics_list, index=model_names,
                   columns=['explained_variance', 'mae', 'mse', 'r2'])  # 建立回归指标的数据框
print('all samples: %d \t features: %d' % (n_samples, n_features),'\n','-'*60)  # 打印输出样本量和特征数量
regresstion_score  # 模型回归指标

all samples: 506 	 features: 13 
 ------------------------------------------------------------

# 模型效果可视化
plt.figure(figsize=(10, 10)) 
for i, pre_y in enumerate(pre_y_list):  
    plt.subplot(len(pre_y_list)+1,1,i+1) # 子图6行*1列
    plt.plot(np.arange(len(y_test)), y_test, color='k', label='true y')  
    plt.plot(np.arange(len(y_test)), pre_y_list[i], 'g--', label=model_names[i])  
    plt.title('True and {} result comparison'.format(model_names[i])) 
    plt.legend(loc='upper right')  
    plt.tight_layout() # 自动调整子图间隔

# 上述初始模型XGBR与GBR表现较优。这里以XGBR为例进行网格搜索+交叉验证
clf = XGBRegressor(random_state=0)  # 建立GradientBoostingRegressor回归对象,该模型较好处理特征量纲与共线性问题
parameters = {
              'n_estimators': [10, 50, 100, 500],
              'learning_rate': [0.05, 0.1, 0.3, 0.5],
              'max_depth': [5, 6, 7, 10]}  # 定义要优化的参数信息
model_gs = GridSearchCV(estimator=clf,
                        param_grid=parameters, cv=5, scoring='r2', n_jobs=-1)  # 建立交叉检验模型对象
model_gs.fit(X_train, y_train)  # 训练交叉检验模型
print('Best score is:', model_gs.best_score_)  # 获得交叉检验模型得出的最优得分
print('Best parameter is:', model_gs.best_params_)  # 获得交叉检验模型得出的最优参数

Best score is: 0.7624423570088324
Best parameter is: {'learning_rate': 0.1, 'max_depth': 5, 'n_estimators': 50}

# 获取最佳训练模型
model_xgbr = model_gs.best_estimator_  # 获得交叉检验模型得出的最优模型对象
pre_y = model_xgbr.predict(X_test)

# 模型评估 优于上次
model_metrics_list = [[m(y_test, pre_y) for m in model_metrics_functions]]  # 回归评估指标列表
regresstion_score = pd.DataFrame(model_metrics_list, index=['model_xgbr'],
                   columns=['explained_variance', 'mae', 'mse', 'r2'])  # 建立回归指标的数据框
regresstion_score  # 模型回归指标

# 模型效果可视化
plt.figure(figsize=(10, 2))  # 创建画布
plt.plot(np.arange(len(y_test)), y_test, color='k', label='true y')  # 画出原始值的曲线
plt.plot(np.arange(len(y_test)), pre_y, 'g--', label='XGBR')  # 画出每条预测结果线
plt.title('True and {} result comparison'.format('XGBR'))  # 标题
plt.legend(loc='upper right')  # 图例位置
plt.tight_layout() # 自动调整子图间隔