二手房数据的分析与挖掘（六）-- 房价预测之模型训练

实验环境

Windows7
Anaconda3
matplotlib
seaborn
pandas
numpy

实验内容

在Jupyter notebook中利用numpy、pandas、matplotlib、seaborn进行模型的训练和优化，为后面预测房价提供算法模型。

实验步骤

一、数据来源

特征处理后的训练数据：https://download.csdn.net/download/qq_35809147/11178563

二、代码解析

import numpy as np
import pandas as pd
import matplotlib
from matplotlib import pyplot as plt
import seaborn as sns

train_data = pd.read_csv(u'H://特征处理_train_data.csv')

train_data.head()

train_data.info()

%matplotlib notebook
sns.set(style='whitegrid', context='notebook')  # 设定样式，还原可用 sns.reset_orig

cols = train_data.columns
cm = np.corrcoef(train_data.values.T)  # 计算相关系数
sns.set(font_scale=1.5)

# 画相关系数矩阵的热点图
hm = sns.heatmap(cm, 
        annot=True, 
        square=True,
        fmt='.2f',
        annot_kws={'size': 15},
        yticklabels=cols,
        xticklabels=cols)
plt.tight_layout()
sns.reset_orig()

sns.reset_orig()

y = train_data.Hand
X = train_data.drop(['Hand'], axis=1)

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=42, test_size=0.2)

from sklearn.neighbors import KNeighborsClassifier
estimator = KNeighborsClassifier()
model = estimator.fit(X_train, y_train)
print("R^2 为{0:.2f}".format(model.score(X_test, y_test)))
predict = model.predict(X_test)
print('准确度为：{0:.2f}'.format(np.mean(y_test==predict)*100))

%matplotlib notebook
#优化模型——设置参数，应为近邻个数对邻近算法影响最大，所以从n_neighbores出发来优化模型
from sklearn.neighbors import KNeighborsClassifier
from sklearn.cross_validation import cross_val_score
all_scores = []
avg_scores =[]
for n_neighbors in range(1,21):
    estimator = KNeighborsClassifier(n_neighbors=n_neighbors)
    scores = cross_val_score(estimator, X = X, y = y)
    all_scores.append(scores)
    avg_scores.append(np.mean(scores))
    

import matplotlib.patches as mpatches
el = mpatches.Ellipse((2.98, 0.86),0.3, 0.4 , angle=30, alpha=0.2)
plt.figure(figsize=(9, 6))
plt.plot(range(1, 21), avg_scores, 'bo-')
plt.xlabel(u'近邻个数')
plt.ylabel(u'效果得分')
#plt.annotate(u'最优值', xy = (2.52, 0.86), xytext=(6.39, 0.86),  arrowprops=dict(arrowstyle="fancy",color="0.5", patchB=el,shrinkB=0,connectionstyle="arc3,rad=0.3"))

#由上图可以得出当n_neighbors的值为1是，得分最高，
max_score = avg_scores[0]
print(u"最优值为：{0:.2f}".format(max_score))

for parameter_value, scores in zip(range(1, 21), all_scores):
    n = len(scores)
    plt.plot([parameter_value]*n, scores, 'bo-')

from sklearn.neighbors import KNeighborsClassifier
estimator = KNeighborsClassifier(n_neighbors=1)
model = estimator.fit(X_train, y_train)
print("R^2 为{0:.2f}".format(model.score(X_test, y_test)))
predict = model.predict(X_test)
print('准确度为：{0:.2f}%'.format(np.mean(y_test==predict)*100))

from sklearn.neighbors import KNeighborsClassifier
estimator = KNeighborsClassifier(n_neighbors=1)
predict_model = estimator.fit(X, y)
#将训练的模型保存到文件当中
from sklearn.externals import joblib
joblib.dump(predict_model, 'd://predict_model.m')

三、训练后的模型下载：

训练后的模型：https://download.csdn.net/download/qq_35809147/11179050

欢迎加入大数据学习交流群，一起交流学习！