Market Basket Analysis Streamlit

marketbasket_streamlit.py

def encode_units(x):
        if x <= 0:
            return 0
        else:
            return 1

def mba(df, support):
    df1 = df.copy()
    df2 = df1[df1.sum(axis=1) > 1]
    frequent_itemsets = apriori(df2.astype('bool'), min_support=support, use_colnames=True)
    rules = association_rules(frequent_itemsets, metric="lift", min_threshold=1).sort_values(by='lift', ascending =False)
    return rules


def frequent_bought_recommender(basket_df, item, support):
    basket_sets = basket_df.copy()
    df_item = basket_sets.loc[basket_sets[item] == 1]
    item_frequent = apriori(df_item, min_support=support, use_colnames=True)
    a_rules = association_rules(item_frequent, metric="lift", min_threshold=1)
    # Sorting on lift and support
    a_rules.sort_values(['lift','support'],ascending=False).reset_index(drop=True)
    recommender = a_rules['consequents'].unique()[:5]
    return recommender