def gen_rules(filepath, args):
if not os.path.exists(filepath):
print(filepath, 'not exists,please set the filepath')
print('\n\n\n')
print(
'------------------------------处理文件%s-----------------------------------'
% (os.path.basename(filepath)))
with open(filepath, encoding='utf-8') as f:
dataSet = [line.split() for line in f.readlines()]
if len(dataSet) <= 100: ##如果交易数据少于100条，返回空
print(
'-----------------------交易数据小于100条，不生成rules-------------------------'
)
return []
n = args.support * len(dataSet)
initSet = fpgrowth.createInitSet(dataSet)
myFPtree, myHeaderTab = fpgrowth.createFPtree(initSet, n)
freqItems = []
fpgrowth.mineFPtree(myFPtree, myHeaderTab, n, set([]), freqItems)
# for x in freqItems:
# print(x)
# compute support values of freqItems
suppData = fpgrowth.calSuppData(myHeaderTab, freqItems, len(dataSet))
suppData[frozenset([])] = 1.0
# for x, v in suppData.items():
# print(x, v)
# freqItems = [frozenset(x) for x in freqItems]
# print(freqItems)
rules = fpgrowth.generateRules(freqItems,
suppData,
minConf=args.confidence)
filter_rules = [rule for rule in rules if len(rule[1]) == 1]
filter_rules = sorted(filter_rules, key=lambda p: p[2], reverse=True)
print('number of association rules:\n', len(filter_rules))
return filter_rules
# myFPtree.disp()
# print fpgrowth.findPrefixPath('z', myHeaderTab)
# print fpgrowth.findPrefixPath('r', myHeaderTab)
# print fpgrowth.findPrefixPath('t', myHeaderTab)
# freqItems = []
# fpgrowth.mineFPtree(myFPtree, myHeaderTab, 3, set([]), freqItems)
# for x in freqItems:
# print x
'''kosarak data'''
start = time.time()
n = 20000
with open("E:\dvancedos\database\webdocs.dat", "rb") as f:
parsedDat = [line.split() for line in f.readlines()]
initSet = fpgrowth.createInitSet(parsedDat)
myFPtree, myHeaderTab = fpgrowth.createFPtree(initSet, n)
freqItems = []
fpgrowth.mineFPtree(myFPtree, myHeaderTab, n, set([]), freqItems)
for x in freqItems:
print(x)
print(time.time() - start, 'sec')
# compute support values of freqItems
suppData = fpgrowth.calSuppData(myHeaderTab, freqItems, len(parsedDat))
suppData[frozenset([])] = 1.0
for x, v in suppData.items():
print(x, v)
freqItems = [frozenset(x) for x in freqItems]
fpgrowth.generateRules(freqItems, suppData)