How to use bandpass filter (say 30-60 day) with daily timeseries data (for 365 days) especially in MATLAB? How many data points are going to be reduced after filtering? Any suggestion would be greatly appreciated.
Hi,
I will give only a hint to one detail of your question.
In general it depends on the type of filter, if the number of data points is reduced.
Filters can be designed in a way that no data points reduction takes place.
Depends also on your application, if you need reduced set of data.
Hi,
I am not sure if I got your question right but if you want to bandpass your data for frequencies where one cycle corresponds to 30 and 60 days and you collected data for 365 days this code is what you need in Matlab. Note that depending on your task a butterworth filter might not be the right/best one.
highcut = (365/30)/(365/2); %define highcut frequency 30 days in radians
lowcut =(365/60)/(365/2); %define lowcut frequency 60 days in radians
[b,a] = butter(6,[lowcut highcut]); %create 6th order butterworth band pass filter
freqz(b,a); %look at your filter characteristics
FilteredData = filtfilt(b,a,DataIn); %filter Data without phase shift
This will not reduce the data points however be careful interpreting filtered signals close to the beginning and ending of the dataset.
I hope this helps!
Thank you very much Niels Niethard . I really appreciate your suggestion and it works well with my data.
Regards,
Satya
Niels Niethard : I need a clarification that why to divide the total number of days by two (e.g., 365/2), while computing highcut and lowcut?
You can't filter frequencies fast then the half of your sampling rate. This is called Nyquist freqency and is always the half of the originally sampled freuqency. In your case 182.5 days instead of 365.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining