def union_features_one(x):
q = set(process_str_replace(x['query']))
t = set(process_str_replace(x['product_title']))
d = set(process_str_replace(x['product_description']))
qt = len(q & t)
qtd = len(q & (t | d))
return pd.Series({'qt': qt, 'qtd': qtd,
'qtn': qt/len(q),
'qtdn': qtd/len(q)})
def union_features_one(x):
q = set(process_str_replace(x['query']))
t = set(process_str_replace(x['product_title']))
d = set(process_str_replace(x['product_description']))
qt = len(q & t)
qtd = len(q & (t | d))
return pd.Series({
'qt': qt,
'qtd': qtd,
'qtn': qt / len(q),
'qtdn': qtd / len(q)
})
def get_str_for_query7(train, q, median_relevance, id=None):
df = train[(train['query']==q) &
(train['median_relevance']==median_relevance) &
(train.index!=id)]
title_set = set()
for e in df['product_names'].values:
title_set |= set(process_str_replace(e))
return title_set
def create_similarity_features7(train, row, id=None):
clr_print(id, row['product_title'])
tx1 = get_str_for_query7(train, row['query'], 1, id)
tx2 = get_str_for_query7(train, row['query'], 2, id)
tx3 = get_str_for_query7(train, row['query'], 3, id)
tx4 = get_str_for_query7(train, row['query'], 4, id)
our_tx = set(process_str_replace(row['product_names']))
return len(tx1 & our_tx), len(tx2 & our_tx), len(tx3 & our_tx), len(tx4 & our_tx)