You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be nice to expose a utils module that contains methods for things like my plot_hist() code in the description of the PR above that can be used to generate plots of our Aggregations.
Similarly, we could provide utils for generating confusion matrix plots, PR curves, and other items related to evaluation. All with native support for FiftyOne data types as input.
Now that FiftyOne works natively in notebooks, leveraging matplotlib for these kind of outputs is at 50% towards native in-App plots (same information, minus the interactively), while also being significantly easier to implement.
The text was updated successfully, but these errors were encountered:
I think we'll get the best flexibility out of plotting if we add the option to output Aggregation results as Pandas Dataframes. We could make Dataframes the default, even.
This was largely addressed in v0.7.3, where a convention was established that things like evaluation results come in an EvaluationResults class, which provides methods for generating various supported plots, etc.
Note that AggregationResult was removed in favor of returning the raw aggregation results, which means that FiftyOne is shirking responsibility for things like plot_hist() (the user has the raw data, so they can do it themselves).
From #777 (comment):
It would be nice to expose a utils module that contains methods for things like my
plot_hist()
code in the description of the PR above that can be used to generate plots of ourAggregation
s.Similarly, we could provide utils for generating confusion matrix plots, PR curves, and other items related to evaluation. All with native support for FiftyOne data types as input.
Now that FiftyOne works natively in notebooks, leveraging matplotlib for these kind of outputs is at 50% towards native in-App plots (same information, minus the interactively), while also being significantly easier to implement.
The text was updated successfully, but these errors were encountered: