Generally, in supervised learning, we have many algorithms. But I am confused to choose which one is good for a problem. how to find that linear models can be used in this case or trees or SVM or KNN in particular case??
Asked
Active
Viewed 9 times
0
-
A general recommendation, try to understand the particular mechanics of the underlying starting generating process (or processes) and its subsequent evolution to create your data. Use this to generate data, for which you know the true parameter values. Then, test various methodologies and compare results based on accuracy. Clueless on this, look at results for various apps and present to experts and get opinions. – AJKOER Jun 28 '20 at 13:48
-
@AJKOER: It could be helpful to add a reference to this approach. – Michael M Jun 28 '20 at 13:51
-
As requested a source example: https://research.aimultiple.com/synthetic-data/ which discusses various aspects of the synthetic-data approach. To quote in part: Agent-based modeling: To achieve synthetic data in this method, a model is created that explains an observed behavior, and then reproduces random data using the same model. It emphasizes understanding the effects of interactions between agents on a system as a whole. – AJKOER Jun 28 '20 at 13:55