Machine learning in trading: theory, models, practice and algo-trading - page 130
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
briefly if anyone is interested....
I tried to use the DTW algorithm in clustering, with simple elementary data just for myself...
I took a price and trained RF on its parts in the form of a sliding window, then I again took the same price and the same parts in the form of a sliding window and clustered the DTW algorithm and then trained RF
The error when training and recognizing new data with the DTW algorithm was 2-4% lower
if you add a large number of predictors then I think the error can be much more lower, but the algorithm is slow as hell
briefly if anyone is interested....
I tried to use the DTW algorithm in clustering, with simple elementary data just for myself...
I took a price and trained RF on its parts in the form of a sliding window, then I again took the same price and the same parts in the form of a sliding window and clustered the DTW algorithm and then trained RF
The error when training and recognizing new data with the DTW algorithm was 2-4% lower
If you add a large number of predictors then I think the error can be lowered a lot more, but the algorithm is slow as hell
Input price... Predictors on input...
Horse-drawn circus!
The price of entry... Predictors at the entrance...
A horse-drawn circus!
Vadim, if you have something to say, say it...
If you just want to say something at the 5th grade level, it's better to be silent...
at least out of respect for those who will read it later...
Vadim, if you have something to say, say it...
And if you just want to make a fifth-grade remark, it's better to remain silent...
At least out of respect for those who will read it later...
Predictor do you know what it is?
Why put a price on the net?
1) Do you know what a predictor is?
2) Why do you need to feed the network a price?
1) well yes, the data on which the network learns
2) and what is more objective than the price? So why not?
briefly if anyone is interested....
I tried to use the DTW algorithm in clustering, with simple elementary data just for myself...
I took a price and trained RF on its parts in the form of a sliding window, then I again took the same price and the same parts in the form of a sliding window and clustered the DTW algorithm and then trained RF
The error when training and recognizing new data with the DTW algorithm was 2-4% lower
if you add a large number of predictors then I think the error can be much lower, but the algorithm is slow as hell
In brief, what is the point of doing the clustering of time series in the context of RF training?
If you want to keep it short, I will answer with a picture...
we have two rows of the same length with the same pattern inside, but one pattern does not exactly match the indexes of the other, so the essence of the DTW algorithm is that when clustering it will understand that this is the same pattern, but RF will not understand
It's all roughly speaking of course.
1) well, yes, the data on which the network learns
2) and what is more objective than price? , so why not?
Success. :)
If you want to keep it short, I will answer with a picture...
we have two rows of the same length with the same pattern inside, but one pattern is not exactly the same index as the other, so the essence of DTW algorithm is that in clustering it will understand that it is the same pattern, but RF will not understand