I use weather data and date data to predict a restaurant turnover, features include: day, month, whether holiday, temperature, weather conditions, wind size, etc., is these characteristics, roughly tried catboost and xgboost respectively, the results of the two methods on the test set is similar, but the quality of these two methods generate importance ranking is different, want to know what are the reasons why, in the consult everybody a great god,

At left is the result of xgboost, right is the result of catboost,

CodePudding user response:

I need to look at your code to determine

CodePudding user response:

You can do your xgboost and catboost code for me? I want to do with R catboost but don't have the package, please advice