def write_predictdb():
cursor.execute(
'''UPDATE UpdateFilm SET linear_predict=%s,linear_test=%s,
lasso_predict=%s,lasso_test=%s,knn_predict=%s,knn_test=%s,poly_predict=%s,poly_test=%s
WHERE imdb_filmID = %s''',
(scoredict['linear_predict'], scoredict['linear_test'],
scoredict['lasso_predict'], scoredict['lasso_test'],
scoredict['knn_predict'], scoredict['knn_test'],
scoredict['poly_predict'], scoredict['poly_test'], filmid))
db.commit()
from db_helper.init_cursor import cursor
from db_helper.init_cursor import db
def get_exist_list():
exist_tup = cursor.fetchall()
exists = list()
for a_item in exist_tup:
exists.append(a_item[0])
return exists
cursor.execute('SELECT tags FROM FilmDB')
res = dict()
tags = get_exist_list()
for dd in tags:
for tag in dd.split('/'):
if tag!='':
res[tag] = res.get(tag, 0) + 1
print res
return int(res)
else:
return 0
def get_weekgross(s):
for item in s:
if item.h4:
if item.h4.string:
if item.h4.string.startswith('Opening Weekend'):
return get_num(item.get_text())
cursor.execute(
'SELECT FilmDB.imdb_filmID '
'FROM FilmDB,TrailerClick '
'WHERE FilmDB.imdb_filmID=TrailerClick.imdb_filmID '
'AND (country=\'USA\'OR country=\'UK\') AND gross>1000000 and openweek_gross is null'
)
filmids = get_exist_list()
print filmids
for filmid in filmids:
soup = page_read.page_read_nolog('http://www.imdb.com/title/' + filmid +
'/')
if soup.select('.txt-block'):
weekgross = get_weekgross(soup.select('.txt-block'))
if weekgross and weekgross != 0:
cursor.execute(
'''UPDATE TrailerClick SET openweek_gross=%s WHERE imdb_filmID=%s''',
(weekgross, filmid))
db.commit()
def write_predictdb():
cursor.execute(
'''UPDATE UpdateFilm SET linear_predict=%s,linear_test=%s,
lasso_predict=%s,lasso_test=%s,knn_predict=%s,knn_test=%s,poly_predict=%s,poly_test=%s
WHERE imdb_filmID = %s''',
(scoredict['linear_predict'], scoredict['linear_test'],
scoredict['lasso_predict'], scoredict['lasso_test'],
scoredict['knn_predict'], scoredict['knn_test'],
scoredict['poly_predict'], scoredict['poly_test'], filmid))
db.commit()
cursor.execute('SELECT click_times,gross '
'FROM FilmDB,TrailerClick '
'WHERE FilmDB.imdb_filmID=TrailerClick.imdb_filmID '
'AND gross>4*TrailerClick.click_times')
clicks = cursor.fetchall()
X_R1 = list()
y_R1 = list()
for click_time, gross in clicks:
X_R1.append(click_time)
y_R1.append(gross)
X_data = np.array(X_R1).reshape(-1, 1)
y_data = np.array(y_R1).reshape(-1, 1)
X_train, X_test, y_train, y_test = train_test_split(X_data,
y_data,
test_size=0.2,
random_state=4)
cursor.execute(
'SELECT TrailerClick.imdb_filmID,max(click_times) FROM UpdateFilm,TrailerClick WHERE TrailerClick.imdb_filmID=UpdateFilm.imdb_filmID GROUP BY UpdateFilm.imdb_filmID'
from db_helper.init_cursor import db
from db_helper.init_cursor import cursor
def get_exist_list():
exist_tup = cursor.fetchall()
exists = list()
for a_item in exist_tup:
exists.append(a_item[0])
return exists
cursor.execute('SELECT imdb_filmID FROM UpdateFilm ')
filmids = get_exist_list()
cursor.execute('SELECT imdb_filmID FROM TrailerClick')
all_ids = get_exist_list()
for a_id in all_ids:
if a_id not in filmids:
cursor.execute('DELETE FROM TrailerClick WHERE imdb_filmID=%s',
(a_id, ))
db.commit()