Closed kimyoungjin06 closed 4 years ago
Dear Kim,
I tried to set the extrema_detection="parabol"
with an hypothetical signal, and it worked just fine for me. If you could share the signal that you are using (i.e., the tidal wave), it would be easier for others to test and possibly help you.
Best
@rfmiotto Thanks! I upload data at my temporary repository.
,date,tidal_level
0,2010-09-02 00:00:00,317.0
1,2010-09-02 01:00:00,271.0
2,2010-09-02 02:00:00,268.0
3,2010-09-02 03:00:00,310.0
4,2010-09-02 04:00:00,381.0
5,2010-09-02 05:00:00,451.0
6,2010-09-02 06:00:00,500.0
...
And the code I used for importing the data.
import pandas as pd
station = stations[0]
df = pd.read_csv('./Data_hour_sorted/%s.csv'%station, index_col=0)
t = pd.to_datetime(df.date).values
s = df.tidal_level.values
win_size = 24*28*2
_s = s[:win_size]
_t = [datetime.utcfromtimestamp(_*1e-9) for _ in t[:win_size].astype(int)] # dt64 to datetime
_t = np.array([time.mktime(x.timetuple()) for x in _t])
After that, same code in original text.
FYI
Hi @kimyoungjin06,
Apologies but it's going to take mea while to test it and get back to you.
The traceback you provided points at scaling line which is based on x
-distance between points. My suspicions are your time-array (_t
) and your systems data formats.
As for helping you debug, I have a couple of suggestions:
_t
. That will remove suspicion from the _s
.Again, apologies for not helping more and quicker. Let me know how your search is going and I'll try to help on the earliest chance.
Dawid
@laszukdawid
Thanks for your kind answer!
I get the result with floats instead of timestamp _t
.
It is so helpful for me.
Have a nice day without COVID-19!
Young Jin Kim
I understand that the problem is now resolved when using floats instead of timestamps. Great to hear that :)
Sorry that it took me sooo long to get back. Crazy times.
Feel free to reopen if the issue isn't closed.
Hope you're also doing well :)
Problem
I try to apply EEMD to a tidal wave.
So, I follow simplest example of EEMD of documentation. The example code is a success, but I get shape error with divide by zero encountered in true_divide in applying with a tidal wave.
Data
An input data is hourly tidal level in 28 days.
CASE 1: without parabolic option
CASE 2: with parabolic option