PythonPandas之间的时间流逝动态范围

Question

我是一个新的python/pandas用户。我试图得到时间序列数据帧的动态范围（基于值的差异）之间的时间增量（以秒为单位）。我的示例数据帧是：

time                          price
2013-04-26 09:30:03-04:00       101
2013-04-26 09:30:04-04:00       101
2013-04-26 09:30:05-04:00       102
2013-04-26 09:30:06-04:00       105
2013-04-26 09:30:07-04:00       104
2013-04-26 09:30:08-04:00       105
2013-04-26 09:30:09-04:00       106
2013-04-26 09:30:10-04:00       104
2013-04-26 09:30:11-04:00       110
2013-04-26 09:30:12-04:00       109
2013-04-26 09:30:13-04:00       111
2013-04-26 09:30:14-04:00       108
2013-04-26 09:30:15-04:00       106
2013-04-26 09:30:16-04:00       107
2013-04-26 09:30:17-04:00       107
2013-04-26 09:30:18-04:00       108
2013-04-26 09:30:19-04:00       109
2013-04-26 09:30:20-04:00       109
2013-04-26 09:30:21-04:00       110

我试图得到价格差为4的时间差。一旦价格差异达到，这个价格点就成为下一次计算的“起点”，以此类推。 所需的结果类似于（以秒为单位的时间增量）：

^{pr2}$

Answer 1

不知道这在性能方面有多好。在

import numpy as np
import pandas as pd

gen = df.price.iteritems()

def get_deltas(gen):
    time, value = next(gen)
    deltas = [np.nan]  # initial value
    for line in gen:
        if np.abs(line[1] - value) >= 4:
            deltas.append(np.abs(line[0] - time))
            time, value = line
        else:
            deltas.append(np.nan)    
    return deltas

deltas = get_deltas(df.price.iteritems())
df['deltas'] = deltas

In [58]: df
Out[58]: 
                     price   deltas
time                               
2013-04-26 13:30:03    101      NaN
2013-04-26 13:30:04    101      NaN
2013-04-26 13:30:05    102      NaN
2013-04-26 13:30:06    105  0:00:03
2013-04-26 13:30:07    104      NaN
2013-04-26 13:30:08    105      NaN
2013-04-26 13:30:09    106      NaN
2013-04-26 13:30:10    104      NaN
2013-04-26 13:30:11    110  0:00:05
2013-04-26 13:30:12    109      NaN
2013-04-26 13:30:13    111      NaN
2013-04-26 13:30:14    108      NaN
2013-04-26 13:30:15    106  0:00:04
2013-04-26 13:30:16    107      NaN
2013-04-26 13:30:17    107      NaN
2013-04-26 13:30:18    108      NaN
2013-04-26 13:30:19    109      NaN
2013-04-26 13:30:20    109      NaN
2013-04-26 13:30:21    110  0:00:06