Conjoint analysis in Python using a Max Diff sample, creating a score and ranking

Question

I am learning how to do some conjoint analysis using a max diff questionnaire. I created some dummy content with some code (probably more complex code than needed, but here it is):

test = pd.DataFrame(
    np.arange(20).reshape(
        20,1)).applymap(
    lambda x: np.random.choice(
        list(['orange', 'apple', 'banana']), size=2,replace=False))
test.columns = ['groups']
test_1 = pd.DataFrame(test['groups'].to_list(), columns=['most', 'least'])

test_2 = pd.DataFrame(
    np.arange(20).reshape(
        20,1)).applymap(
    lambda x: np.random.choice(
        list(['banana', 'pear', 'peach']), size=2,replace=False))
test_2.columns = ['groups']
test_2 = pd.DataFrame(test_2['groups'].to_list(), columns=['most', 'least'])

test_3 = pd.DataFrame(
    np.arange(20).reshape(
        20,1)).applymap(
    lambda x: np.random.choice(
        list(['apple', 'banana', 'pear']), size=2,replace=False))
test_3.columns = ['groups']
test_3 = pd.DataFrame(test_3['groups'].to_list(), columns=['most_3', 'least_3'])

df = test_1.join(test_2, lsuffix='_1', rsuffix='_2').join(test_3, lsuffix='_3')

The code should you give a dataframe with data that looks like this:

user	most_1	least_1	most_2	least_2	most_3	least_3
0	orange	banana	pear	peach	apple	banana
1	orange	banana	peach	banana	pear	banana
2	orange	banana	pear	peach	apple	banana
3	orange	banana	pear	banana	apple	banana
4	banana	apple	peach	pear	banana	apple
5	orange	banana	pear	banana	banana	apple
6	banana	orange	pear	peach	banana	pear
7	apple	banana	peach	banana	pear	banana
8	orange	apple	pear	banana	banana	pear
9	orange	apple	banana	pear	apple	banana
10	apple	banana	pear	banana	pear	apple
11	apple	banana	pear	banana	apple	banana
12	apple	banana	banana	peach	banana	apple
13	orange	banana	peach	pear	pear	banana
14	apple	banana	banana	peach	apple	banana
15	apple	banana	peach	banana	banana	apple
16	apple	orange	peach	banana	pear	banana
17	banana	apple	pear	banana	apple	pear
18	apple	orange	peach	banana	pear	banana
19	banana	apple	peach	banana	pear	apple

So in this example, a person should choose their favorite fruit between an orange, apple, and banana. The next question would be banana, peach, and pear. Final one is apple, banana, and pear. It includes more questions, but I stopped at 3 for the example.

I am trying to create new columns for each fruit in the dataframe. Each fruit will have a point column (+1 if in most, -1 in least, which I have the code for, see below) and a rank column (most points equal 1, least equals 5). In the rank column, if two fruits have the same point value, it will compare them when they were in the same group. For example, if apple and banana have the same score, the fruit that did best when they were in the same group (1 and 3) will be ranked higher. The idea would be I can now see which fruit a specific user prefers and in what order.

So the new table will have this added:

|user | orange_pt | banana_pt | apple_pt | pear_pt | peach_pt | orange_rank | banana_rank | apple_rank | pear_rank | peach_rank |:------|:------:|:------:|:------:|:------:| |:------:|:------:|:------:|:------:|:------:|:-----:| 0|1|-2|1|1|-1|1|5|2|3|4

(not sure why the formatting is not working here)

This should be repeated for each user (row) in the dataframe.

Here is the code for scoring:

df.join((df.loc[:,df.columns.str.contains('most')] \
.apply(pd.value_counts, axis=1).fillna(0).astype(int) - \
df.loc[:,df.columns.str.contains('least')] \
.apply(pd.value_counts,axis=1).fillna(0). \
astype(int)).add_suffix('_score'))

So I really just need to figure out a way to create a ranking. Any help greatly appreciated!

THANKS!!!