Moving window of 100 points with 20 points overlap.
1 visualización (últimos 30 días)
Mostrar comentarios más antiguos
davit petraasya
el 15 de Feb. de 2016
Comentada: Image Analyst
el 16 de Feb. de 2016
Hi everybody
I have around 500 points. My x axis-times, y-magnitudes. I need to do plot moving window of 100 points averages with 20 point overlap over time. How I may do the plot? I attached something sample plot also. (Please see my code and excel file, I need to continue my this code ...
[NUMERIC,TXT,RAW]=XLSREAD('sihex.xlsx');
A= -0.59;
B = 43.45;
C=0.09;
x=NUMERIC(:,2);
y=NUMERIC(:,1);
Mag=NUMERIC(:,3);
for iii=1:length(TXT);
date_str(iii,:)=TXT(iii,1);
end
date_num=datenum(date_str,'dd/mm/yyyy');
condition11=(Mag >= 0.8 & x > A-C & x < A+C & y > B-C & y < B+C);
plot(date_num(condition11),Mag(condition11),'.b');
datetick('x','yyyy','keeplimits'))
0 comentarios
Respuesta aceptada
dpb
el 15 de Feb. de 2016
Nfilt=20;
b=ones(1,Nfilt)/Nfilt; % weights
magFiltered=[nan filter(b,1,mag)]; % weight average (includes end effects)
magFiltered(2:nFilt-1)=nan; % replace end effects w/ NaN keeping time location
3 comentarios
dpb
el 15 de Feb. de 2016
I've no klew what the "100 events" means, the above is a moving average of nPoints.
Oh, rereading I missed part of the request...no, the above doesn't do the "jump", it's a step increment of unity. As IA says, use either blockproc or conv instead.
Más respuestas (1)
Image Analyst
el 15 de Feb. de 2016
An alternate way is to use blockproc(), if you have the Image Processing Toolbox. It gives you control over the window size and the "jump" length.
Or, you could use conv() (which moves over one element at a time) and then just subsample every 20th element from the result.
2 comentarios
Image Analyst
el 16 de Feb. de 2016
I've attached two demos where I use the function in a wide variety of ways. Adapt as needed. Good luck.
Ver también
Categorías
Más información sobre Line Plots en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!