# Predict compatibility of 2 people as boolean classification problem

How can I predict the compatibility of 2 people as a boolean classification problem?

I want to know if below is an appropriate approach to modelling compatibility, or if I should be using "market basket analysis" or some other approach instead?

I'm less interested in the specific result below, and more interested in if this is a realistic way to frame this data science problem.

Background:

Assume people only have 3 attributes: compassion, extroversion and humor. These are also boolean and can be modelled as 1s and 0s in a list ([compassion, extroversion, humor]).

So someone with all 3 characteristics would be [1,1,1] and someone with only humor would be [0,0,1].

We have pairs of people who match and do not match, specified by 1 or 0, where 1=match and 0=no_match.

How to solve this?

I don't consider this a simple distance problem (ie: euclidean distance) because its very possible that generally an introvert and extrovert get along, but 2 extroverts don't.

Data:

person1   person2     match?
--------  --------    ------
([1,1,0], [1,0,1]) => 1
([0,0,0], [1,1,1]) => 1
([1,0,1], [1,0,0]) => 0
([1,1,1], [0,1,0]) => 0
([0,0,0], [0,1,1]) => 1
([1,1,0], [1,1,1]) => 0
([1,0,0], [1,0,1]) => 0
([0,0,1], [0,0,0]) => 0
([0,0,0], [0,0,1]) => 0
([0,0,0], [0,1,1]) => 1
([0,1,0], [0,1,1]) => 0
([0,1,0], [0,1,1]) => 0
([0,1,0], [1,0,0]) => 1


What I've tried:

My first thought was to concatenate both individuals' data for each example. Then use that to fit the model.

Data structured as python code:

X_train = [
[1,1,0,1,0,1],
[0,0,0,1,1,1],
[1,0,1,1,0,0],
[1,1,1,0,1,0],
[0,0,0,0,1,1],
[1,1,0,1,1,1],
[1,0,0,1,0,1],
[0,0,1,0,0,0],
[0,0,0,0,0,1],
[0,0,0,0,1,1],
[0,1,0,0,1,1],
[0,1,0,0,1,1],
[0,1,0,1,0,0],
]

y_train = [1,1,0,0,1,0,0,0,0,1,0,0,1]

X_test = [
[0,1,1,0,0,0],
[1,1,0,1,0,1],
[1,0,0,1,0,0],
[0,0,0,1,0,0],
[0,1,0,0,0,0],
[0,0,0,0,0,0],
]

y_test = [1,1,0,0,1,0]


Computing the match:

from sklearn.metrics import classification_report

from sklearn.ensemble import RandomForestClassifier
rf = RandomForestClassifier()
rf.fit(X_train, y_train)
y_pred = rf.predict(X_test)

print(classification_report(y_test,y_pred))

precision    recall  f1-score   support

0       0.60      1.00      0.75         3
1       1.00      0.33      0.50         3

avg / total       0.80      0.67      0.62         6


I'm less concerned about the results here and more interested in if this is a proper way to frame this problem.

Can you offer a suggestion?

• I don't understand the question completely - are you asking whether Random Forests are a good choice or do you want to model your data differently? Sep 26 '18 at 7:49
• @André I want to know if this is an appropriate approach to modelling compatibility, or if I should be using "market basket analysis" or some other approach instead? Sep 26 '18 at 14:30