有一个数据框包含以下信息检测器id, channel id(每个检测器都有一些 channel )和时间戳(为简单起见,让它是一个整数)和给定(detector_id, channel_id) 对。
如何计算给定 (detector_id, channel_id) 对中最后一个非零事件后经过的天数?
这是一个例子:
df = pd.DataFrame({
"time": [1, 1, 2, 3, 3, 4, 4],
"detector_id": [0, 0, 0, 0, 0, 0, 1],
"channel_id": [0, 0, 1, 0, 1, 1, 1],
"counts": [0, 1, 0, 1, 0, 1, 0],
})
我尝试通过以下方式解决这个问题:
df["diff"] = df["time"] - df.groupby(["detector_id", "channel_id"])['time'].diff()
它产生以下结果:
time detector_id channel_id counts diff expected
0 1 0 1 0 NaN NaN
1 2 0 1 1 1.0 NaN
2 3 0 1 0 2.0 1.0
3 4 0 0 1 NaN NaN
4 5 0 1 0 3.0 3.0
5 6 0 1 1 5.0 4.0
6 7 1 1 0 NaN NaN
如您所见,给定的解决方案没有考虑 counts 列。一旦我们看到 counts > 0,我们应该将差异设置为零,否则传播。
请您参考如下方法:
这应该很接近,但需要对您的完整数据进行测试:
def f(subdf):
ffilled = (subdf.loc[subdf['counts'] > 0, 'time']
.reindex_like(subdf)
.ffill()
.shift())
return subdf['time'] - ffilled
df['diff'] = (df.groupby(['detector_id', 'channel_id'])
.apply(f)
.sort_index(level=-1)
.values)
time detector_id channel_id counts diff expected
0 1 0 1 0 NaN NaN
1 2 0 1 1 NaN NaN
2 3 0 1 0 1.0 1.0
3 4 0 0 1 NaN NaN
4 5 0 1 0 3.0 3.0
5 6 0 1 1 4.0 4.0
6 7 1 1 0 NaN NaN
