ML之回歸預(yù)測(cè)：利用十(xgboost,10-1)種機(jī)器學(xué)習(xí)算法對(duì)無(wú)人駕駛汽車系統(tǒng)參數(shù)(2017年的data,18+2)進(jìn)行回歸預(yù)測(cè)值VS真實(shí)值——bug調(diào)試記錄

?

?

?

目錄

輸出結(jié)果

1、增加X(jué)GBR算法

---

?

?

?

?

輸出結(jié)果

1、增加X(jué)GBR算法

1、增加X(jué)GBR算法時(shí)候，采用網(wǎng)格搜索的方法

XGBR_grid_model Training time： 135.60037931849538 輸出XGBR_grid_model模型的最優(yōu)參數(shù)： {'learning_rate': 0.03, 'max_depth': 4, 'n_estimators': 100} XGBR_grid_model_best_score: 0.7993051604810518 XGBR_grid_model_score： -0.339632045000555

2、增加X(jué)GBR算法時(shí)候，調(diào)用得到的最佳參數(shù)，卻輸不出最佳參數(shù)對(duì)應(yīng)的準(zhǔn)確度！

?

3、增加X(jué)GBR算法時(shí)候，對(duì)新數(shù)據(jù)進(jìn)行預(yù)測(cè)，遇到bug：因?yàn)閄GBR算法，要求傳入的數(shù)據(jù)都為數(shù)值型

成功解決ValueError: DataFrame.dtypes for label must be int, float or bool

?

4、默認(rèn)配置參數(shù)，輸出的準(zhǔn)確度較低：54%
(1)、通過(guò)特征選擇可得到最高準(zhǔn)確度為67.69%

ML之回歸預(yù)測(cè)：利用十種機(jī)器學(xué)習(xí)(LiR+xgboost(特征重要性+特征選擇))算法對(duì)無(wú)人駕駛汽車系統(tǒng)參數(shù)(2017年的data,18+2)進(jìn)行回歸預(yù)測(cè)值VS真實(shí)值

?

XGBR：The value of default parameter of XGBR is 0.539021005879261 XGBR：R-squared value of DecisionTreeRegressor: 0.539021005879261 XGBR：測(cè)試[12.8, 13.0]行數(shù)據(jù)， [53.609283 53.47569 53.920986 53.920986 53.61446 53.61446 53.82615753.773552 53.44194 53.09419 54.089035 54.353115 53.321796 53.29324353.970306 53.752117 53.460567 53.44237 53.840027 53.920986 52.3508557.61133 57.598843 57.911274 58.06042 57.945023 57.665913] ------------------------------------------------------------ XGBR_model.feature_importances_： [0.08985916 0.01444405 0.08940411 0. 0.03163605 0.018708770.00869713 0.12159647 0.03933521 0.12161936 0.02289704 0.042608730.02663714 0.04179822 0.03441375 0.14182347 0.01409451 0.14042689] Thresh=0.000, n=18, Accuracy: 52.56% Thresh=0.009, n=17, Accuracy: 52.56% Thresh=0.014, n=16, Accuracy: 51.94% Thresh=0.014, n=15, Accuracy: 52.89% Thresh=0.019, n=14, Accuracy: 54.23% Thresh=0.023, n=13, Accuracy: 53.38% Thresh=0.027, n=12, Accuracy: 52.86% Thresh=0.032, n=11, Accuracy: 53.08% Thresh=0.034, n=10, Accuracy: 64.67% Thresh=0.039, n=9, Accuracy: 67.69% Thresh=0.042, n=8, Accuracy: 66.25% Thresh=0.043, n=7, Accuracy: 67.52% Thresh=0.089, n=6, Accuracy: 64.64% Thresh=0.090, n=5, Accuracy: 40.05% Thresh=0.122, n=4, Accuracy: 37.93% Thresh=0.122, n=3, Accuracy: 19.23% Thresh=0.140, n=2, Accuracy: 22.15% Thresh=0.142, n=1, Accuracy: 11.77%

?

?

?

?

?

?

?

?

?

?

?

?

?

?

總結(jié)

以上是生活随笔為你收集整理的ML之回归预测：利用十(xgboost,10-1)种机器学习算法对无人驾驶汽车系统参数(2017年的data,18+2)进行回归预测值VS真实值——bug调试记录的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問(wèn)題。

如果覺(jué)得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。