def js_wp_xsd(xlz, yh_xsd_dd=False):
dy_rq, dy_jx = js_dy_rq(xlz)
print('-' * 80)
print('Generate: items co-rated similarity matrix')
dy_xsd_sz = {}
dydyh_sj = cssj(dayinbushu=1000)
for yh, dys in xlz.items():
for dy in dys:
dy_xsd_sz.setdefault(dy, defaultdict(int))
for dy2 in dys:
if dy == dy2:
continue
if yh_xsd_dd:
dy_xsd_sz[dy][dy2] += 1 / math.log(1 + len(dys))
else:
dy_xsd_sz[dy][dy2] += 1
dydyh_sj.js_sj()
print('Generate:items co-rated similarity matrix.(Done)')
print('-' * 80)
dydyh_sj.js()
print('-' * 80)
print('Calculate: item-item similarity matrix')
dy_xs_sz_sj = cssj(dayinbushu=1000)
for dy, zj_wp in dy_xsd_sz.items():
dy1_cd = dy_rq[dy]
for dy2, jx in zj_wp.items():
yh2_cd = dy_rq[dy2]
dy_xsd_sz[dy][dy2] = jx / math.sqrt(dy1_cd * yh2_cd)
dy_xs_sz_sj.js_sj()
print('Calculate:item-item similarity matrix(Done)')
print('-' * 80)
dy_xs_sz_sj.js()
return dy_xsd_sz, dy_rq, dy_jx
def jx_yh_xsd(xlsjb, yh_csb_xsd=False):
print('-' * 80)
print('Build: movie-users inverse table')
dydyh = collections.defaultdict(set)
dyymd = defaultdict(int)
for yh, dys in xlsjb.items():
for dy in dys:
dydyh[dy].add(yh)
dyymd[dy] += 1
print('Build: movie-users inverse table(Done)')
print('-' * 80)
dy_zh = len(dydyh)
print('total movie number = %d' % dy_zh)
print('-' * 80)
print('Generate:user co-rated movies similarity matrix')
yhxsd_sz = {}
dydyh_sj = cssj(dayinbushu=1000)
for dy, yhm in dydyh.items():
for yh1 in yhm:
yhxsd_sz.setdefault(yh1, defaultdict(int))
for yh2 in yhm:
if yh1 == yh2:
continue
if yh_csb_xsd:
yhxsd_sz[yh1][yh2] += 1 / math.log(1 + len(yhm))
else:
yhxsd_sz[yh1][yh2] += 1
dydyh_sj.js_sj()
print('Generate: user co-rated movies similarity matrix(Done)')
print('-' * 80)
dydyh_sj.js()
print('-' * 80)
print('Calculate matrix: user-user similarity ')
yh_xsd_sz = cssj(dayinbushu=1000)
for yh1, zj_yh in yhxsd_sz.items():
yh1_cd = len(xlsjb[yh1])
for yh2, js in zj_yh.items():
yh2_cd = len(xlsjb[yh2])
yhxsd_sz[yh1][yh2] = js / math.sqrt(yh1_cd * yh2_cd)
yh_xsd_sz.js_sj()
print('Calculate: user-user similarity(Done)')
print('-' * 80)
yh_xsd_sz.js()
return yhxsd_sz, dyymd, dy_zh
def yuce(self, csj):
dy_tj = defaultdict(list)
print('-' * 80)
print('start to predict scores ')
yc_sj = csh.cssj(dayinbushu=500)
for gd, yh in enumerate(csj):
jl_sj = self.tuijian(yh)
dy_tj[yh].append(jl_sj)
yc_sj.js_sj()
print('Predict scores(Done)')
print('-' * 80)
yc_sj.js()
return dy_tj
def yuce(self, csj):
dytuijian = defaultdict(list)
print('-' * 80)
print('start to predict scores ')
ybyshijian = cssj(dayinbushu=500)
for ni, yh in enumerate(csj):
jl_dy = self.tuijian(yh)
dytuijian[yh].append(jl_dy)
ybyshijian.js_sj()
print('Predict scores(Done)')
print('-' * 80)
ybyshijian.js()
return dytuijian
def cs(self, csj):
if not self.n_jl_dy or not self.xlj or not self.dy_rqz or not self.dy_hj:
raise ValueError('Method has not called yet.')
self.csj = csj
print('-' * 80)
print('Start to test recommendation system ')
weizhi = self.n_jl_dy
jida = 0
jl_hj = 0
cs_hj = 0
sy_jl_dy = set()
rqzs = 0
cs_sj = cssj(dayinbushu=1000)
for w, yh in enumerate(self.xlj):
cs_dy = self.csj.get(yh, {})
jl_dy = self.tuijian(yh)
for dy in jl_dy:
if dy in cs_dy:
jida += 1
sy_jl_dy.add(dy)
rqzs += math.log(1 + self.dy_rqz[dy])
jl_hj += weizhi
cs_hj += len(cs_dy)
cs_sj.js_sj()
qxd = jida / (1.0 * jl_hj)
hs = jida / (1.0 * cs_hj)
fg = len(sy_jl_dy) / (1.0 * self.dy_hj)
zmd = rqzs / (1.0 * jl_hj)
f_s=(qxd*hs*2.0)/(qxd+hs)
print('Run test recommendation system(Done)')
print('-' * 80)
cs_sj.js()
print('-' * 90)
print('precision=%.4f\trecall=%.4f\tF-Measure1=%.4f\tcoverage=%.4f\tpopularity=%.4f\n' %
(qxd, hs,f_s, fg, zmd))
print('-' * 90)
def cs(self, csj):
if not self.n_jilu_dy or not self.xlji or not self.dy_zmd or not self.dy_jx:
raise ValueError('Method has not called yet.')
self.ceshiji = csj
print('-' * 80)
print('Start:Test recommendation system ')
ni = self.n_jilu_dy
jida = 0
jilu_jihe = 0
ceshi_jihe = 0
sy_jl_dy = set()
rq_hj = 0
cs_sj = csh.cssj(dayinbushu=1000)
for gd, yh in enumerate(self.xlji):
cs_dy = self.ceshiji.get(yh, {})
jl_dy = self.tuijian(yh)
for dy in jl_dy:
if dy in cs_dy:
jida += 1
sy_jl_dy.add(dy)
rq_hj += math.log(1 + self.dy_zmd[dy])
jilu_jihe += ni
ceshi_jihe += len(cs_dy)
cs_sj.js_sj()
xmd = jida / (1.0 * jilu_jihe)
zhaohui = jida / (1.0 * ceshi_jihe)
zy = len(sy_jl_dy) / (1.0 * self.dy_jx)
rqz = rq_hj / (1.0 * jilu_jihe)
f_s = (xmd * zhaohui * 2.0) / (xmd + zhaohui)
print('Run test recommendation system(Done)')
cs_sj.js()
print('-' * 90)
print(
'precision=%.4f\trecall=%.4f\tF-Measure1=%.4f\tcoverage=%.4f\tpopularity=%.4f\n'
% (xmd, zhaohui, f_s, zy, rqz))
print('-' * 90)
def cs(self, csj):
if not self.n_jl_dy or not self.xlj or not self.dy_zmd or not self.dy_hj:
raise ValueError('Method has not called yet.')
self.csj = csj
print('-' * 80)
print('Start to test recommendation system ')
weizhi = self.n_jl_dy
ddjd = 0
jl_zs = 0
cs_zs = 0
sy_ji_zs = set()
rq_hj = 0
ceshishijian = cssj(dayinbushu=1000)
for ni, yh in enumerate(self.xlj):
ceshidy = self.csj.get(yh, {})
jl_dy = self.tuijian(yh)
for dy in jl_dy:
if dy in ceshidy:
ddjd += 1
sy_ji_zs.add(dy)
rq_hj += math.log(1 + self.dy_zmd[dy])
jl_zs += weizhi
cs_zs += len(ceshidy)
ceshishijian.js_sj()
jqz = ddjd / (1.0 * jl_zs)
sh = ddjd / (1.0 * cs_zs)
fg = len(sy_ji_zs) / (1.0 * self.dy_hj)
zmrq = rq_hj / (1.0 * jl_zs)
f_s = (jqz * sh * 2.0) / (jqz + sh)
print('Test recommendation system(Done)')
ceshishijian.js()
print('-' * 90)
print(
'precision=%.4f\trecall=%.4f\tF-Measure1=%.4f\tcoverage=%.4f\tpopularity=%.4f\n'
% (jqz, sh, f_s, fg, zmrq))
print('-' * 90)
print(str(zhi) +":" + dy_ditu[xiangmu])
def jianli_dy_dt(wenjian, shuju):
jieuo = {}
with open(wenjian, encoding='unicode_escape') as sjl:
hangs = sjl.readlines()
for hang in hangs:
hdz = hang.strip().split(shuju)
jieuo[hdz[0]] = hdz[1]
return jieuo
if __name__ == '__main__':
xt_sj = cssj(danzi="Main Function")
print('-' * 80)
print('\t GU JIAKAI 20161795\tWEN JIALI 20174950\t LIU JIAQI 20173122')
print('-' * 80 + '\n')
print('\t\tRecommend System for Movielens ')
print('-' * 70)
print('\t 1:ml-100k\t2:ml-1m\t3:ml-25m')
print('-' * 70 + '\n')
a = int(input("Select Data set:"))
sjj_mc = None
mx_mc = None
if a == 1:
sjj_mc = 'ml-100k'
elif a == 2:
sjj_mc = 'ml-1m'