-
Notifications
You must be signed in to change notification settings - Fork 1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Feature] GridSearchCV for Surprise algorithms #6
Comments
@NicolasHug, Do you confirm that this is the way to go or is there any other alternative that I am not aware of? Thanks, |
Hey Maher, No you don't need to use RDT, you can simply build the docs locally. If you haven't already, install all the packages for development with
this will install sphinx (the documentation generator), the RTD theme and some sphinx extensions needed in the doc building process. The you should be able to build the docs locally by running
You can check the results in I'll add all this in the |
Hello @NicolasHug, I have finished the implementation and basic documentation of the feature, and pushed the changes to my repo here. When you can please take a look at the docs to confirm that the functionality is as you expected. In addition, I want to write a user-guide page but I need to take your opinion on where to add it. Happy new year btw :) Maher |
This is awesome! For the doc, you can put it as the first subsection of the "Advanced Usage", right before "Manually iterate over folds". As this is a major feature, you can also add a quick example in the README under the "example" section (if you don't no problem, I'll do it ^^) I haven't thoroughly tested the code for now (I'll come back to you, probably on Monday), so these are just a few remarks after taking a quick glance at it:
Also, please create a pull request so that we can review changes on the pull request . Your upcoming commits will be automatically added to the pull request discussion. Thank you so much for putting time into it, this will be a very attractive feature! Happy new year :)! Nicolas |
Hey, coming back to you on this :) It all seem to work very well, I only have a few more remarks:
And I think this is pretty much it :) Nicolas |
Hello Nicolas, I have created the pull request #7 . I have also added a user guide in the beginning of the advanced usage section. Please find my comments on each of the points above:
Notes:
Looking forward for the merge ;) Maher |
Awesome! I put my remarks on the pull request. |
Closing this as it has been solved by 714be0b |
param_grid = {'n_epochs': [5, 10], 'lr_all': [0.002, 0.005], Cross-validation using grid searchgrid_search = GridSearchCV(SVD, param_grid, measures=['rmse', 'mae'], cv=3, n_jobs=1) Find the best parameters on the data setdata = Dataset.load_builtin('ml-100k') TypeError Traceback (most recent call last) TypeError: init() got an unexpected keyword argument 'measures' |
Add the ability for a user to easily test an algorithm with different parameters similar to GridSearchCV of sklearn.
One can give an algorithm a dictionary of the different parameters to try and it generates the best combination of parameters based on some error measurements. Since the data model of Surprise does not directly plug into GridSearchCV an analogous functionality should be added.
The text was updated successfully, but these errors were encountered: