Given different combinations of four feature,
a, b, c, d, we can get different different sets of labels from a fixed label set
x, y, z, e.g.
a + b=> Labels
a + b + d=> Labels
x + y
c + d=> Labels
a + b + c + d=> Labels
x + y + z
Not all permutations of the features lead to one or more labels.
To put it into a little more intuitive context, imagine if we have 4 ingredients (features) and we have 3 meals (labels). Combinations of the ingredients will result in 0 to 3 labels, e.g.
- ingredients = ['egg', 'bread', 'steak', 'lettuce']
- labels = ['breakfast', 'lunch', 'dinner']
And we have combinations like:
egg + bread=> Labels
egg + lettuce + steak=> Labels
lunch + dinner
egg + bread + steak=> Labels
breakfast + dinner
egg + bread + lettuce=> Labels
breakfast + lunch
What are some valid data visualization tools can we use to present the combinations of features and resulting labels? esp when
i) presenting all possible preset/predefined combinations ii) users pick the features/ingredients (maybe through a checkbox form) and we display the appropriate label
We can use these but they're not elegant or I can foresee how they'll look like
- a decision tree, unless all branches are binary and leaf nodes can have repeats
- a table where the rows represent the different combinations of the labels and columns represent the features and we use checkmarks on the value cells to indicate that given the checked columns we'll result in the appropriate row label.