In Market basket analysis, lift represents the increase in the sale of item A when you sell item B. Lift is calculated as the confidence of (A, B) divided by the support of B. If the lift value is ‘n’, the intuition behind lift is the likelihood of buying both items A and B is n times more if the items are non-correlated.
Lift (A=>B) = 1 represents there is no correlation between the items
Lift (A=>B) > 1 represents there is positive correlation between the items and there is more likelihood to buy item A when item is bought.
Lift (A=>B) < 1 represents there is a negative correlation between items there is less likelihood to buy item A when item B is bought.
You can sign up for these Data Science online courses by Intellipaat to learn Market basket analysis, recommendation systems and other Machine learning algorithms.
You can watch this video on Market basket analysis to know more about lift:
Back