if start == True:
data[-1] = row.CompleteTimestamp
else:
data = list()
data += [
temp_caseid, row.Activity, row.Resource, '', row.Amount,
row.CompleteTimestamp
]
start = False
table.append(data)
headers = [
'CASE_ID', 'Activity', 'Resource', 'StartTimestamp', 'Amount',
'CompleteTimestamp'
]
df = pd.DataFrame(table, columns=headers)
eventlog = Eventlog(df)
eventlog = eventlog.assign_timestamp(name='StartTimestamp',
new_name='StartTimestamp',
_format='%Y.%m.%d %H:%M:%S',
errors='raise')
eventlog = eventlog.assign_timestamp(name='CompleteTimestamp',
new_name='CompleteTimestamp',
_format='%Y/%m/%d %H:%M:%S',
errors='raise')
eventlog['Duration'] = (eventlog['CompleteTimestamp'] -
eventlog['StartTimestamp']).apply(to_minute)
eventlog.dropna(subset=['Resource', 'StartTimestamp', 'CompleteTimestamp'],
inplace=True)
eventlog.to_csv('../sample_data/BPIC2012.csv')
eventlog = eventlog.loc[eventlog['Resource'].isin(valid_resource_list)]
#20회 초과 이벤트 제외
event_count = eventlog.groupby('CASE_ID').Activity.count()
invalid_case = event_count[event_count > 20].astype(int)
invalid_case_list = list(invalid_case.index)
eventlog = eventlog.loc[~eventlog['CASE_ID'].isin(invalid_case_list)]
#특정 날짜(scheduling 대상)에 event가 있는 모든 instance 추출
eventlog['StartDate'] = eventlog['StartTimestamp'].dt.date
eventlog['CompleteDate'] = eventlog['CompleteTimestamp'].dt.date
d='2012-03-10'
target_date=pd.to_datetime(d).date()
valid_case = eventlog.loc[eventlog['StartDate']==target_date,'CASE_ID'].unique()
eventlog = eventlog.loc[eventlog['CASE_ID'].isin(valid_case)]
#target date 이후 이벤트 삭제 (하루치만)
eventlog = eventlog.loc[eventlog['StartDate']<=target_date]
#weight 배정
max_amount = eventlog['Amount'].max()
print(max_amount)
#custom_bucket_array = np.linspace(0, max_amount, 10)
labels = [x+1 for x in range(5)]
eventlog['weight'] = pd.cut(df['Amount'], 5, labels=labels)
eventlog.to_csv('../result/modi_BPI_2012_0301.csv')
print(eventlog)
print(len(eventlog['CASE_ID'].unique()))
