Machine learning in trading: theory, models, practice and algo-trading - page 1855

New comment
 
Mihail Marchukajtes:
Come on. Let's see what the beast is :-)

well, we all like specialists here, not man-skill

 
Mihail Marchukajtes:
Come on. Let's see what the beast is :-)


At the end of the vid did not divide it into a track and a test, I was exhausted in the evening.

Here are the errors

>>> clf_rf.score(X_train, y_train.values.ravel())
1.0
>>> clf_rf.score(X_test, y_test.values.ravel())
0.8395061728395061
 
Maxim Dmitrievsky:


At the end of the video did not divide into a track and a test, steamed up in the evening.

Here are the errors

Could you briefly describe the essence of the experiment? And its results? I don't really want to spend 40 minutes on something that can be read in 3 minutes.
You can post it on the blog, and then link it here. Then it's easier to find interesting experiments in blogs than here among 19 thousand entries. That's what I've been doing lately.
 
elibrarius:
Could you briefly describe the experiment? And its results? I do not really want to spend 40 minutes on something that can be read in 3 minutes.
It is possible in the blog, and here a link. Then it's easier to find interesting experiments in blogs than here among 19 thousand entries.

With 25 minutes of the essence

 
Maxim Dmitrievsky:


At the end of the video did not divide into a track and a test, steamed up in the evening.

Here are the errors.

What are you feeding as signs to the forest? There can not be zero error..., maybe you just put the current cluster number in the signs? explicitly or not explicitly?

 
mytarmailS:

What are you giving as signs in the forest? there can not be zero error..., maybe you just threw the number of the current cluster in the signs? explicitly or not explicitly?

if you are doing this "by chance" after a couple of years, i feel sorry for you )) some kind of childish mistakes, not even childish, but dumb

in the last example with increment clustering, on catbust there was a 0 error on oos, this is normal

 
Maxim Dmitrievsky:

if you are doing this after a couple of years, i feel sorry for you )) childish mistakes, not even childish, but daunting

in the last example with clustering of increments, there was a 0 error on the oos on catbust, this is normal

Don't give me sympathy, it's daunting to answer questions asked with a lefty line. Make a TS on new data with your algorithm, we'll feel sorry for you all together when the bug comes out.

 
mytarmailS:

No need to sympathize with me, it's downright stupid to answer a question with a left-hand line. Make a TS on new data with your algorithm, we'll all sympathize with you when the bug comes out.

I showed you the technology of making a TS via clustering, the first step. the rest of the dumbass suggestions you implement on your own

 
Maxim Dmitrievsky:

I showed the technology of making ts through clustering, the first step. the rest of the dumb suggestions implement your own

Now I will implement your super technology, let's see what happens.

 
mytarmailS:

Now I'm going to implement your super technology, see what happens...

You'll be surprised if you do it right.

1...184818491850185118521853185418551856185718581859186018611862...3475
New comment
Reason: