Remove some allocations

[rubdos]

Performance improvements, i.e. #260 / #261

Fixes no specific issue

Checklist

• My branch is up-to-date with development branch.
• Everything works and tested on latest stable Rust.
• Coverage and Linting have been applied

Current behaviour

smartcore svm is a bit slow

New expected behaviour

smartcore svm is a bit less slow! About 4% faster training in my very simple 5 features x 159 samples test.

Change logs

This removes some allocations by reusing vectors and clearing them in iterations.

[codecov-commenter]

Codecov Report

Merging #262 (59c5993) into development (9cd7348) will increase coverage by 0.02%.
The diff coverage is 41.93%.

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

@@               Coverage Diff               @@
##           development     #262      +/-   ##
===============================================
+ Coverage        45.38%   45.40%   +0.02%     
===============================================
  Files               85       85              
  Lines             7231     7232       +1     
===============================================
+ Hits              3282     3284       +2     
+ Misses            3949     3948       -1     

Impacted Files	Coverage Δ
src/linalg/basic/vector.rs	50.00% <0.00%> (-5.56%)	⬇️
src/svm/svc.rs	35.66% <47.82%> (+0.88%)	⬆️
src/svm/svr.rs	46.03% <57.14%> (+0.39%)	⬆️

... and 13 files with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[rubdos]

Second commit makes training another 14% faster on my test data.

[Mec-iS]

thanks a lot, i will look into it as soon as possible. Please run rustfmt src/*.rs to avoid the linting test failing.

[rubdos]

FWIW, rustfmt does not complain about my changes; those lines are unrelated to this PR. I'll gladly run rustfmt if you still want me to, however.

I currently notice that my accuracy estimation (possibly train_test_split) is the slowest part in my test code. I'm playing with KNN now, however (Model trained in 1921 ms; accuracy 0.9921170101624086 (estimated in 189165 ms)).

The difference between training and estimation is only train_test_split and ClassificationMetricsOrd::accuracy().get_score...

[Mec-iS]

good job!

Tomorrow I will have some time at the computer to take a closer look to your contribution.

clippy is complaining for unnecessary parenthesis. For the sake of making the tests pass please apply the requested minor changes even if present in other files.

Thanks again for your time spent using Smartcore.

[morenol]

I can create a PR fixing clippy in a couple of hours. That is probably related to the new lints added to the stable version released a couple of days ago

[Mec-iS]

thanks Luis 👍
obviously we want 100% to stick with clippy standards.

[morenol]

PR opened at: #263

We can merge that first. Then we can rebase these changes. Hopefully, CI will be green after that

[morenol]

@rubdos could you rebase your changes? It seems that there are not conflicts

[morenol]

good catch!

[morenol]

looks good to me. But needs a rebase

[Mec-iS]

Thanks a lot for this.

If you rebase your branch with current development I will merge.

[rubdos]

If you rebase your branch with current development I will merge.

Done!