This repository was archived by the owner on Feb 28, 2024. It is now read-only.
This repository was archived by the owner on Feb 28, 2024. It is now read-only.
plot_evaluations() incompatible with BayesSearchCV #1056
Description
I ran multiple models with BayesSearchCV, and try to plot the optimization results, plot_histogram and plot_objective both work perfectly fine, however plot_evaluations is throwing this error every time:
AttributeError: 'Real' object has no attribute 'categories'
And shows the first few figures where there are categorical variables, and empty axes elsewhere.
The complete traceback:
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-63-f0c2a2bf3c15> in <module>
5 clf_params = list(filter(is_clf_param, model_res.best_params_.keys()))
6 n_params = len(clf_params)
----> 7 plot_evaluations(model_res.optimizer_results_[0])
8 fig, axs = plt.subplots(1, n_params, figsize=(5*n_params,4))
9 for i,ax in enumerate(axs):
/opt/conda/lib/python3.7/site-packages/skopt/plots.py in plot_evaluations(result, bins, dimensions, plot_dims)
829 index, dim = plot_dims[i]
830 if iscat[j]:
--> 831 bins_ = len(dim.categories)
832 elif dim.prior == 'log-uniform':
833 low, high = space.bounds[index]
AttributeError: 'Real' object has no attribute 'categories'