问题描述：数据处理，尤其是遇到大量数据且需要for循环处理时，需要消耗大量时间，如代码1所示。通过data['trip_time'][i]的方式会占用大量的时间

代码1

import time
t0=time.time()
for i in range(0,len(data.index)):
    data['trip_time'][i] = pd.Timestamp(data['lpep_dropoff_datetime'][i]) - pd.Timestamp(data['lpep_pickup_datetime'][i])
t1=time.time()
print(t1 - t0)

解决办法，添加.at定位索引，data.at[i,'trip_time']

import time
t0=time.time()
for i in range(0,len(data.index)):
    data.at[i,'trip_time'] = pd.Timestamp(data.at[i,'lpep_dropoff_datetime']) - pd.Timestamp(data.at[i,'lpep_pickup_datetime'])
t1=time.time()
print(t1 - t0)

 

评价：可以看出 使用at进行索引的方法相比loc、iloc和ix要快了将近1000倍！

%timeit outdf.loc[0] = indf.loc[0]
100 loops, best of 3: 11.7 ms per loop
%timeit outdf.iloc[0] = indf.iloc[0]
100 loops, best of 3: 11.4 ms per loop
 %timeit outdf.ix[0] = indf.ix[0]
100 loops, best of 3: 11.6 ms per loop
%timeit outdf.at[0,'time'] = indf.at[0,'time']
10000 loops, best of 3: 25.3 µs per loop