Python+sklearn使用线性回归算法预测儿童身高
In [1]:
import copyimport numpy as npfrom sklearn import linear_modeldef linearRegressionPredict(x, y): lr = linear_model.LinearRegression() # 拟合 lr.fit(x, y) return lr# 儿童年龄,性别（0女1男）,父亲身高,母亲身高,祖父身高,祖母身高,外祖父身高,外祖母身高x = np.array([[1, 0, 180, 165, 175, 165, 170, 165],\ [3, 0, 180, 165, 175, 165, 173, 165],\ [4, 0, 180, 165, 175, 165, 170, 165],\ [6, 0, 180, 165, 175, 165, 170, 165],\ [8, 1, 180, 165, 175, 167, 170, 165],\ [10, 0, 180, 166, 175, 165, 170, 165],\ [11, 0, 180, 165, 175, 165, 170, 165],\ [12, 0, 180, 165, 175, 165, 170, 165],\ [13, 1, 180, 165, 175, 165, 170, 165],\ [14, 0, 180, 165, 175, 165, 170, 165],\ [17, 0, 170, 165, 175, 165, 170, 165]])# 儿童身高，单位：cmy = np.array([60, 90, 100, 110,\ 130, 140, 150, 164,\ 160, 163, 168])# 根据已知数据拟合最佳直线的系数和截距lr = linearRegressionPredict(x, y)# 查看最佳拟合系数print('k:', lr.coef_) k: [ 8.03076923e+00 -4.65384615e+00 2.87769231e+00 -5.61538462e-01 1.33226763e-15 5.07692308e+00 1.88461538e+00 0.00000000e+00]
In [2]:
# 截距print('b:', lr.intercept_) b: -1523.1538461538462
In [3]:
import matplotlib.pyplot as plt
In [23]:
x.take(0,axis =1)
Out[23]:
array([ 1, 3, 4, 6, 8, 10, 11, 12, 13, 14, 17])
In [24]:
plt.plot(x.take(0,axis =1),y)
Out[24]:
[]
In [23]:
# 预测xs = np.array([[10, 0, 180, 165, 175, 165, 170, 165],\ [17, 1, 173, 153, 175, 161, 170, 161],\ [34, 0, 170, 165, 170, 165, 170, 165]])for item in xs: # 深复制，假设超过18岁以后就不再长高了 item1 = copy.deepcopy(item) if item1[0] > 18: item1[0] = 18 print(item, ':', lr.predict(item1.reshape(1,-1)))
[ 10 0 180 165 175 165 170 165] : [140.56153846][ 17 1 173 153 175 161 170 161] : [158.41][ 34 0 170 165 170 165 170 165] : [176.03076923]
In [35]:
x[:][0]
Out[35]:
array([ 1, 0, 180, 165, 175, 165, 170, 165])