Virgil Chan

I can confirm that the same error occurs when using the example: ```python y_true = [3, -0.5, 2, 7] y_pred = [2.5, 0.0, 2, 8] ``` I will create a...

> My naive guess would be whoever implemented it saw the log and needed to ensure the log was not of a negative number. I'm surprised it was never an...

@sgiardl Thank you for reporting the bug. The pull request has been merged, and it should be resolved in the next update.

> Concerning the CI failure, see #31722. I think this is fixed in `main` by #31690. So updating the branch should work.

Probably the simplest fix would be to update the documentation to inform end-users about the change in dtype. For example: "Note: When applied to a pandas DataFrame, `SimpleImputer` may change...

Commenting to stay updated!

It appears that the temperature scaling neural network is defined by the function: $$\left[ y_1, \cdots, y_n \right] \mapsto \mathrm{softmax}\left( \left[ \frac{y_1}{T}, \cdots, \frac{y_n}{T} \right] \right),$$ where $\displaystyle \sum_{i=1}^n y_i...

Could we make it solvable by constraining ourselves with the "matrix scaling" case (and hence the "temperature" and "vector" cases)? It also seems like we have fewer arguments to worry...

Thank you for the replies; they are really helpful! I see there are two points of interest: 1. Decide if the calibration methods improve the original predictions in a generic...

Hello, I've implemented a simple version of temperature scaling with `numpy` and `scipy` using the `_temperature_scaling` function. The function takes the following arguments: - `predictions`: `(n_samples, n_classes)` array, the output...