Remove a part of data and replace it with data to fit trend

11 views (last 30 days)
VISHNURAJ SREERAJ NAIR
VISHNURAJ SREERAJ NAIR on 9 Jan 2022
Commented: Hiro on 11 Jan 2022
Hello, can some one please help me with the removal of data in the photo. I tried all the outlier removals, smooth data and conventional funtions, but it is unable to detect.

Answers (1)

Hiro
Hiro on 10 Jan 2022
Try this. It removes the dent you see in the plot. Please note this also affect the both edges of the profile.
[cleanedData,outlierIndices,thresholdLow,thresholdHigh] = filloutliers(slip1,...
"linear","movmean",29000,"ThresholdFactor",1,"SamplePoints",T1);
It would also be a good idea to detrend first and then find the outliers by using isoutlier.
  1 Comment
Hiro
Hiro on 11 Jan 2022
I found a better solution.
  1. decompose the profile into long-term, short-term and roughness profiles.
  2. detect outliers in roughness and replace them with other values (linear interpolation in this case)
  3. re-compose the new roughness back to the long and short profiles together
Here's the code:
[LT, ST, R] = trenddecomp(slip1);
plot(T1,LT,T1,ST,T1,R);
[Rnew,outlierIndices,thresholdLow,thresholdHigh] = filloutliers(R,...
"linear","quartiles","SamplePoints",T1);
slip1New = Rnew + LT + ST;
plot(T1,spring1,T1,slip1New);

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by