根据device_id+ds+order_id为唯一标识给片段打标签
df1['temp']=0 #temp记录与上一条的不同数
for i in range(len(df1)):
if (df1.device_id.shift()[i]!= df1.device_id[i]) or (df1.ds.diff()[i]!=0) or (df1.rw.diff()[i]!=0):
df1.temp[i]+=1
df1['label']=df1.temp[0] #label是累积数,也是片段的唯一标识
for i in range(len(df1)):
df1.label[i]=sum(df1.temp[0:(i+1)])
#print(i) #查看循环轮数
#shift是一阶差分函数