AdaBoost Reg

AI & Machine learning
1

hi,

I made AdaBoost regression and AdaBoost classification model and  but after that, I had no idea :

1- is there any possibility to see the model by Graphviz or any thing else

2- how to use it to analyze the next day return (here we have 10 features in the code, I attached the code)

I need some hints or any references on that, please.



Thanks

2

Hi Amir,



I looked through your notebook and I found some things you can improve. 

  1. First, you are using 10 features but all of those features are highly correlated. You should have features that have a low correlation between them.
  2. You are using this code "data.ret1.rolling(window=5).sum()" to calculate returns of 5 days I suppose. If you want percentage returns over any number of days you can use "pct_change()" directly. Like in this case you could've used it like "pct_change(5)".
  3. You are checking the accuracy of your predicted value which are the percentage returns of the next day. I should warn you if you test the accuracy of predicted percentage returns it won't be very high. Because the regressor won't be able to predict the returns up to such a high degree of accuracy. For example, it might predict the returns as 2%, but the actual returns are 2.2%. You can make a custom function that checks the accuracy of predicted values with some level of variance.


Yes, it is possible to visualise models using Graphviz. You can follow this link, it will guide you on how to do so.