fitrtree takes forever to run in MATLAB 2022a

Dear MATLAB team,

About two years ago, I used to fit a decision tree regression (fitrtree) to a huge feature matrix (e.g, 5,000,000 rows by 20 columns) in MATLAB 2018b and MATLAB 2020b. The operation often took less than 5 minutes to execute (I don't actually remember the exact duration though, but it was really fast for such a huge data).

However, this is not the case in later versions like MATLAB 2022a; the computation never finishes. I even converted the input arguments into tall arrays, thinking parallel computing would help. The code was still running for more than 8 hours; so I had to cancel it and migrate it to MATLAB 2018b which executed it in 65 seconds. You can check it out with MATLAB 2023a to see if the same problem exists.

I would be grateful if you could help me out. And if this is a technical problem, kindly fix it for us quickly.

Regards

Answer 1

[Raymond Norris]

Hi @Richard Fiifi Annan. I would suggest you contact MathWorks Technical Support to look into this.