ML之LiR：利用LiR线性回归算法(自定义目标函数MSE和优化器GD)对Boston房价数据集(两特征+归一化)进行回归预测
目录
利用LiR线性回归算法(自定义目标函数MSE和优化器GD)对Boston房价数据集(两特征+归一化)进行回归预测
设计思路
输出结果
核心代码
相关文章
ML之LiR：利用LiR线性回归算法(自定义目标函数MSE和优化器GD)对Boston房价数据集(两特征+归一化)进行回归预测
ML之LiR：利用LiR线性回归算法(自定义目标函数MSE和优化器GD)对Boston房价数据集(两特征+归一化)进行回归预测实现
利用LiR线性回归算法(自定义目标函数MSE和优化器GD)对Boston房价数据集(两特征+归一化)进行回归预测
设计思路
输出结果
train_boston_data.shape (1460, 81)Id MSSubClass MSZoning ... SaleType SaleCondition SalePrice0 160 RL ... WD Normal 2085001 220 RL ... WD Normal 1815002 360 RL ... WD Normal 2235003 470 RL ... WD Abnorml 1400004 560 RL ... WD Normal 250000[5 rows x 81 columns]train_t.head() LotFrontage GarageArea SalePrice0 65.0 5482085001 80.0 4601815002 68.0 6082235003 60.0 6421400004 84.0 836250000after scale,train_t.head() LotFrontage GarageArea SalePrice00.207668 0.386460 0.27615910.255591 0.324401 0.24039720.217252 0.428773 0.29602630.191693 0.452750 0.18543040.268371 0.589563 0.331126LotFrontage GarageArea00.207668 0.38646010.255591 0.32440120.217252 0.42877330.191693 0.45275040.268371 0.589563Id MSSubClass LotFrontage ... MoSold YrSold SalePriceId 1.000000 0.011156 -0.010601 ... 0.021172 0.000712 -0.021917MSSubClass0.011156 1.000000 -0.386347 ... -0.013585 -0.021407 -0.084284LotFrontage -0.010601 -0.3863471.000000 ... 0.011200 0.007450 0.351799LotArea -0.033226 -0.1397810.426095 ... 0.001205 -0.014261 0.263843OverallQual -0.028365 0.0326280.251646 ... 0.070815 -0.027347 0.790982OverallCond 0.012609 -0.059316 -0.059213 ... -0.003511 0.043950 -0.077856YearBuilt-0.012713 0.0278500.123349 ... 0.012398 -0.013618 0.522897YearRemodAdd -0.021998 0.0405810.088866 ... 0.021490 0.035743 0.507101MasVnrArea -0.050298 0.0229360.193458 ... -0.005965 -0.008201 0.477493BsmtFinSF1 -0.005024 -0.0698360.233633 ... -0.015727 0.014359 0.386420BsmtFinSF2 -0.005968 -0.0656490.049900 ... -0.015211 0.031706 -0.011378BsmtUnfSF-0.007940 -0.1407590.132644 ... 0.034888 -0.041258 0.214479TotalBsmtSF -0.015415 -0.2385180.392075 ... 0.013196 -0.014969 0.6135811stFlrSF 0.010496 -0.2517580.457181 ... 0.031372 -0.013604 0.6058522ndFlrSF 0.005590 0.3078860.080177 ... 0.035164 -0.028700 0.319334LowQualFinSF -0.044230 0.0464740.038469 ... -0.022174 -0.028921 -0.025606GrLivArea0.008273 0.0748530.402797 ... 0.050240 -0.036526 0.708624BsmtFullBath 0.002289 0.0034910.100949 ... -0.025361 0.067049 0.227122BsmtHalfBath -0.05 -0.002333 -0.007234 ... 0.032873 -0.046524 -0.016844FullBath 0.005587 0.1316080.198769 ... 0.055872 -0.019669 0.560664HalfBath 0.006784 0.1773540.053532 ... -0.009050 -0.010269 0.284108BedroomAbvGr 0.037719 -0.0234380.263170 ... 0.046544 -0.036014 0.168213KitchenAbvGr 0.002951 0.281721 -0.006069 ... 0.026589 0.031687 -0.135907TotRmsAbvGrd 0.027239 0.0403800.352096 ... 0.036907 -0.034516 0.533723Fireplaces -0.019772 -0.0455690.266639 ... 0.046357 -0.024096 0.466929GarageYrBlt 0.000072 0.0850720.070250 ... 0.005337 -0.001014 0.486362GarageCars0.016570 -0.0401100.285691 ... 0.040522 -0.039117 0.640409GarageArea0.017634 -0.0986720.344997 ... 0.027974 -0.027378 0.623431WoodDeckSF -0.029643 -0.0125790.088521 ... 0.021011 0.022270 0.324413OpenPorchSF -0.000477 -0.0061000.151972 ... 0.071255 -0.057619 0.315856EnclosedPorch 0.002889 -0.0120370.010700 ... -0.028887 -0.009916 -0.1285783SsnPorch-0.046635 -0.0438250.070029 ... 0.029474 0.018645 0.044584ScreenPorch 0.001330 -0.0260300.041383 ... 0.023217 0.010694 0.111447PoolArea 0.057044 0.0082830.206167 ... -0.033737 -0.059689 0.092404MiscVal -0.006242 -0.0076830.003368 ... -0.006495 0.004906 -0.021190MoSold 0.021172 -0.0135850.011200 ... 1.000000 -0.145721 0.046432YrSold 0.000712 -0.0214070.007450 ... -0.145721 1.000000 -0.028923SalePrice-0.021917 -0.0842840.351799 ... 0.046432 -0.028923 1.000000[38 rows x 38 columns]coef and intercept: [0.21627565 0.41024884] 0.0543428481373919cost after log： -3.850369422061899 -4.52343070892457best w1 and w2 after GD： 0.10003438525600654 0.30004957896248946
核心代码
LiR = linear_model.LinearRegression()LiR.fit(X_train, y_train)print('coef and intercept: ',LiR.coef_,LiR.intercept_)def CalCostByW(train_df,slope):w1_lists=[];w2_lists=[];cost_lists=[]for i in range (30):for j in range(30):w1= slope*i+0.1w2= slope*j+0.3w1_lists.append(w1); w2_lists.append(w2)cost_lists.append(cost(train_df,train_df.LotFrontage,train_df.GarageArea,w1,w2))# print (cost(train_df))return w1_lists,w2_lists,cost_listsw1_lists,w2_lists,cost_lists=CalCostByW(train_t,0.01)
ML之LiR：利用LiR线性回归算法(自定义目标函数MSE和优化器GD)对Boston房价数据集(两特征+归一化)进行回归预测
