机器学习之LR实现线性预测

线性回归预测房价

任务：
基于usa_housing_price.csv数据，建立线性回归模型，预测合理房价

• 以sqft_living为输入变量，建立单因子模型，评估模型表现，可视化线性回归预测结果
• 以sqft_living、sqft_lot、sqft_above、yr_built、lat为输入变量，建立多因子模型，评估模型表现
• 预测sqft_living=1180、sqft_lot=5650、sqft_above=1180、yr_built=1955、lat=47.5112的合理房价

# load the data
import pandas as pd
import numpy as np
data = pd.read_csv('usa_housing_price.csv')
data.head()

	price	sqft_living	sqft_lot	sqft_above	yr_built	lat
0	221900.0	1180	5650	1180	1955	47.5112
1	538000.0	2570	7242	2170	1951	47.7210
2	180000.0	770	10000	770	1933	47.7379
3	604000.0	1960	5000	1050	1965	47.5208
4	510000.0	1680	8080	1680	1987	47.6168

%matplotlib inline
from matplotlib import pyplot as plt
fig = plt.figure(figsize = (10,10))
fig1 = plt.subplot(231)
plt.scatter(data.loc[:,'sqft_lot'],data.loc[:,'price'])
plt.title('price vs sqft_lot')

fig2 = plt.subplot(232)
plt.scatter(data.loc[:,'sqft_living'],data.loc[:,'price'])
plt.title('price vs sqft_living')

fig3 = plt.subplot(233)
plt.scatter(data.loc[:,'sqft_above'],data.loc[:,'price'])
plt.title('price vs sqft_above')

fig4 = plt.subplot(234)
plt.scatter(data.loc[:,'yr_built'],data.loc[:,'price'])
plt.title('price vs sqft_living')

fig5 = plt.subplot(235)
plt.scatter(data.loc[:,'lat'],data.loc[:,'price'])
plt.title('price vs lat')



plt.show()

#define X and y
X = data.loc[:,'sqft_living']
y = data.loc[:,'price']
y.head()

0    221900.0
1    538000.0
2    180000.0
3    604000.0
4    510000.0
Name: price, dtype: float64

X = np.array(X).reshape(-1,1)
print(X.shape)

(21613, 1)

# set up the linear regression model
from sklearn.linear_model import LinearRegression
LR1 = LinearRegression()
# train the model
LR1.fit(X,y)

LinearRegression()

# Calculate the price vs sqft_living
y_predict_1 = LR1.predict(X)
print(y_predict_1)

[287484.29258296 677805.59158496 172353.54971186 ... 242555.22219424
 405423.10235335 242555.22219424]

# evaluate the model
from sklearn.metrics import mean_squared_error,r2_score
mean_squared_error_1 = mean_squared_error(y,y_predict_1)
r2_score_1 = r2_score(y,y_predict_1)
print(mean_squared_error_1,r2_score_1)

68437189845.45986 0.4928653865220143

fig6 = plt.figure(figsize=(8,5))
plt.scatter(X,y)
plt.plot(X,y_predict_1,'r')
plt.show()

# define X_multi
X_multi = data.drop(['price'],axis=1)
X_multi

	sqft_living	sqft_lot	sqft_above	yr_built	lat
0	1180	5650	1180	1955	47.5112
1	2570	7242	2170	1951	47.7210
2	770	10000	770	1933	47.7379
3	1960	5000	1050	1965	47.5208
4	1680	8080	1680	1987	47.6168
...	...	...	...	...	...
21608	1530	1131	1530	2009	47.6993
21609	2310	5813	2310	2014	47.5107
21610	1020	1350	1020	2009	47.5944
21611	1600	2388	1600	2004	47.5345
21612	1020	1076	1020	2008	47.5941

21613 rows × 5 columns

# set up 2nd linear model
LR_multi = LinearRegression()
#train the model
LR_multi.fit(X_multi,y)

LinearRegression()

# make prediction
y_predict_multi = LR_multi.predict(X_multi)
print(y_predict_multi)

[279081.29418243 832377.00440778 346082.19661749 ... 176687.80346459
 325324.21431742 178529.30559747]

mean_squared_error_multi = mean_squared_error(y,y_predict_multi)
r2_score_multi = r2_score(y,y_predict_multi)
print(mean_squared_error_multi,r2_score_multi)

55814504752.37977 0.5864022564634701

fig7 = plt.figure(figsize=(8,5))
plt.scatter(y,y_predict_multi)
plt.show()

X_test = [1180,5650,1180,1955,47.5112]
X_test = np.array(X_test).reshape(1,-1)
print(X_test)

[[1180.     5650.     1180.     1955.       47.5112]]

y_test_predict = LR_multi.predict(X_test)
print(y_test_predict)

[279081.29418243]