implementing recommender

[Hame-daani]

to-do's

• [x] what features i need?
• [x] what algorithm each feature need?
• [x] what data each algorithm need?
• [x] reimplement recommender using django models
• [x] fix migraion problem with 5000th user
• [x] delete database and reapply migration
• [x] #30
• [x] #32

[Hame-daani]

algorithms

def sim_pearson(p1,p2):
    ...
    return int(similarity)

def top_matches(person):
    ...
    sims = [(sim_pearson(person,friend), friend) for friend in person.friends]
    return sims.sort(reverse=True)[:10]

def get_recommendtion(person):
    ...
    sim*other_rate
    return ranking.sort(reverse=True)[:10]

[Hame-daani]

using django models

example

• similarity model many to many on self? or many to many model with source and target

[Hame-daani]

https://www.youtube.com/watch?v=1xtrIEwY_zY