def dumpPickles(id):
"""
Persist memory intensive data structures, item similarities.
"""
print 'Pickle id %s' % (id)
foldername = '../kfold_cluster_' + id + '/'
print foldername
cluster = algorithm.createPrefs(folder=foldername)
for c in range(8):
prefs = cluster[c]
itemsim = algorithm.calculateSimilarItems(prefs, n=50)
dumpname = "../pickle/" + id + "/oitemsim_" + str(c) + ".pickle"
pickle.dump(itemsim, open(dumpname, "wb"))
def dumpPickles(id):
"""
Persist memory intensive data structures, item similarities.
"""
print 'Pickle id %s'%(id)
foldername = '../kfold_cluster_' + id + '/'
print foldername
cluster = algorithm.createPrefs(folder=foldername)
for c in range(8):
prefs = cluster[c]
itemsim = algorithm.calculateSimilarItems(prefs, n=50)
dumpname = "../pickle/" +id + "/oitemsim_" + str(c) + ".pickle"
pickle.dump(itemsim, open(dumpname, "wb"))
print removed_items
print "Length of new_prefs %d"%(len(new_prefs))
print 'Doing CBR...'
sparsity.caseone_doFillUsingCBR(new_prefs, deleted_users)
print 'Doing Average Filling...'
sparsity.caseone_doAverageFilling(new_prefs, deleted_users)
#use the original prefs here.
print 'Calculating original item similarity...'
oitemsim = original.calculateSimilarItems(org_prefs, n=50)
#use the modified prefs here.
print 'Calculating algorithm item similarity...'
itemsim = algorithm.calculateSimilarItems(new_prefs, n=50)
print 'Calculating original ranking...'
oranking = original.getRecommendedItems(org_prefs, oitemsim,str(user))
print 'Calculating algorithm ranking...'
ranking = original.getRecommendedItems(new_prefs,itemsim,str(user))
print 'Calculating MAE for original...'
original_MAE.append(calMAE(oranking,str(user)))
print 'Calculating MAE for algorithm...'
algorithm_MAE.append(calMAE(ranking,str(user)))
print original_MAE
print removed_items
print "Length of new_prefs %d" % (len(new_prefs))
print 'Doing CBR...'
sparsity.caseone_doFillUsingCBR(new_prefs, deleted_users)
print 'Doing Average Filling...'
sparsity.caseone_doAverageFilling(new_prefs, deleted_users)
#use the original prefs here.
print 'Calculating original item similarity...'
oitemsim = original.calculateSimilarItems(org_prefs, n=50)
#use the modified prefs here.
print 'Calculating algorithm item similarity...'
itemsim = algorithm.calculateSimilarItems(new_prefs, n=50)
print 'Calculating original ranking...'
oranking = original.getRecommendedItems(org_prefs, oitemsim, str(user))
print 'Calculating algorithm ranking...'
ranking = original.getRecommendedItems(new_prefs, itemsim, str(user))
print 'Calculating MAE for original...'
original_MAE.append(calMAE(oranking, str(user)))
print 'Calculating MAE for algorithm...'
algorithm_MAE.append(calMAE(ranking, str(user)))
print original_MAE
print algorithm_MAE
caMAE = []
for cid in range(8):
print "Cluster id: %d" % (cid)
if len(colist) != 0 and len(calist) != 0:
coMAE.append(sum(colist) / len(colist))
caMAE.append(sum(calist) / len(calist))
else:
colist = []
calist = []
clusterPrefs = cluster[cid]
istart = time.time()
itemsim = algorithm.calculateSimilarItems(clusterPrefs, n=50)
istop = time.time()
for user in clusterPrefs:
#fix missing values
if not user:
continue
print "Cluster: %d User: %s" % (cid, user)
rstart = time.time()
#ranking = algorithm.getRecommendedItems(testingPrefs, clusterPrefs,itemsim,user)
ranking = original.getRecommendedItems(testingPrefs, itemsim, user)
rstop = time.time()