特征选择（如何解释结果）？

Question

Let's say i have a dataset and i want to select features which correspond to prediction result better than others. I have implemented some Feature Ranking tests and here are the results:

For prediction model I selected features with best "Mean" value.

X = oil_10[['Sidetrack Code','Well Type Code','Well Status  
Code','Producing Formation','Water Produced, bbl','County']]

Here is the prediction model result with "Best chosen features":

RandomForestRegressor
0.390502562474

And here is the result of prediction model with all dataset features without any selection:

RandomForestRegressor
0.741878611892

How to use Feature Ranking results to implement best prediction result?

Answer 1

好吧，我试图用这种方式解决我的问题：我只删除了最不重要的功能（这些功能的平均重要性值小于0.15），而准确度仍保持75％，但是预测模型现在的运行速度更快。