kaggle房价预测 第四次练习总结（其他知识）

kaggle房价预测参考danB

链接：https://www.kaggle.com/learn/machine-learning

首先是局部依赖图partial dependence plot：简单说就是其他特征边缘化，看这一两个特征的结果特征的依赖关系

import pandas as pd
import matplotlib.pyplot as plt
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_absolute_error
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import Imputer
from xgboost import XGBRegressor
from sklearn.ensemble.partial_dependence import partial_dependence, plot_partial_dependence
from sklearn.ensemble import GradientBoostingRegressor
original_data = pd.read_csv( 'D:/NOTEBOOK/train.csv') #读取训练数据
test_data = pd.read_csv( 'D:/NOTEBOOK/test.csv')  #读取测试数据

original_data_y = original_data.SalePrice #获取y
original_data = original_data.drop(['SalePrice'], axis=1) #删除y

X_train = original_data.select_dtypes(exclude=['object']) #只使用数字预测器
X_test = test_data.select_dtypes(exclude=['object'])

my_imputer = Imputer()
train_X = my_imputer.fit_transform(X_train)
test_X = my_imputer.transform(X_test)
#########################################################################################################
my_model = GradientBoostingRegressor()
my_model.fit(train_X, original_data_y)#变量最好一两个，下面是one-way和two-way
my_plots = plot_partial_dependence(my_model,       
                                   features=[0, 1, (0, 1)], # column numbers of plots we want to show
                                   X=train_X,            # raw predictors data.
                                   feature_names=[ 'BedroomAbvGr','YearBuilt'], # labels on graphs
                                   grid_resolution=10) # number of values to plot on x axis
plt.show()

接下来是Pipelines，简单说把模型拼一起，就有流水线的感觉了。代码后面加上交叉验证，我是分成了5份验证

import pandas as pd
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_absolute_error
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import Imputer
from xgboost import XGBRegressor
from sklearn.pipeline import make_pipeline
from sklearn.model_selection import cross_val_score
original_data = pd.read_csv( 'D:/NOTEBOOK/train.csv') #读取训练数据
test_data = pd.read_csv( 'D:/NOTEBOOK/test.csv')  #读取测试数据

original_data_y = original_data.SalePrice #获取y
original_data = original_data.drop(['SalePrice'], axis=1) #删除y

X_train = original_data.select_dtypes(exclude=['object']) #只使用数字预测器
X_test = test_data.select_dtypes(exclude=['object'])
###############################################################################################
my_pipeline = make_pipeline(Imputer(), RandomForestRegressor())

my_pipeline.fit(X_train, original_data_y)
predictions = my_pipeline.predict(X_test)
################################################################################################
scores = cross_val_score(my_pipeline, X_train, original_data_y, scoring='neg_mean_absolute_error', cv = 5) #交叉验证
print(scores)
print('Mean Absolute Error %2f' %(-1 * scores.mean()))  #取平均值

最好是数据泄露问题，感觉还没理解清楚，等理解清楚再改

个人理解：

1.有的特征是跟着结果变的，所以不考虑，是因为结果才产生的

2.还有就是这个特征是否能用来使用，具体看特征是什么