I've tried predictions with the original danger ratings dataset, which holds the maximum value for Above Treeline, At Treeline, and Below Treeline as its Danger Rating. Try predictions using this new danger ratings dataset with each of these individual columns as Y instead.
This could probably(?) be done rather simply by loading the full dataset and the new avalanche danger dataset, and replacing the danger ratings column in the full dataset with each individual danger rating column in the new danger dataset in turn.
Weather is analyzed to come up with a prediction for each of these elevation areas. For example, if the freezing level is high and there is a lot of precipitation, then higher elevations (Above Treeline) will have a lot of snow which increases avalanche potential while lower elevations (Below Treeline) might receive this as rain which doesn't affect the avalanche danger as much. By using the maximum danger rating of the three, I may have inadvertently made our training data less valid.
I've tried predictions with the original danger ratings dataset, which holds the maximum value for Above Treeline, At Treeline, and Below Treeline as its Danger Rating. Try predictions using this new danger ratings dataset with each of these individual columns as Y instead.
This could probably(?) be done rather simply by loading the full dataset and the new avalanche danger dataset, and replacing the danger ratings column in the full dataset with each individual danger rating column in the new danger dataset in turn.
Weather is analyzed to come up with a prediction for each of these elevation areas. For example, if the freezing level is high and there is a lot of precipitation, then higher elevations (Above Treeline) will have a lot of snow which increases avalanche potential while lower elevations (Below Treeline) might receive this as rain which doesn't affect the avalanche danger as much. By using the maximum danger rating of the three, I may have inadvertently made our training data less valid.