Hi all,
I have a 800 feature model that will like to compress by reducing the number of features and of course trying to maintain the accuracy levels as much as I can.
Is there an algorithm or a technique that I can implement over the features and found the ones that have the highest level of influence over the classes before training a model?
Have tough of training the model over 50 features each time and extract the best ones from every run but this process require a lot of computational resources and was wondering if there's a better way?
Adam W
There are many different ways to do this depending on the model. You could use PCA to transform the features into a set of orthonormal vectors - though sometimes the small variation components are very predictive if your features are market data. You can do Lasso Regression, or use ML models that allow for dimensionality reduction like forests or NNs.
Alejandro Holguin M
Thanks a lot!!!! A lot of work to do
Alejandro Holguin M
The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.
To unlock posting to the community forums please complete at least 30% of Boot Camp.
You can continue your Boot Camp training progress from the terminal. We hope to see you in the community soon!