How to remove outliers without using filter?

5 次查看（过去 30 天）
Nik Rocky
Nik Rocky 2020-12-14
评论： KALYAN ACHARJYA 2020-12-14
Hello,
I have a measure results and there are some random outliers with a big negative or positive values (jitter of uC), there are just random, single values, outliers have a similar value.
So, I want to eliminate this values (just take the same value from left or linear interpolation value ((k-1)+(k+1))/2).
My question is, how to detect this values without using medfilt or filloutliers?
In this case outlier is by motor 3, at 48. sec (negative peak).
Array is in attachement. Thank you!
Yellow peak at 4,8 sek
If I use:
FX = gradient(v3);
I get:
0,209265176669710
0,236204049673113
0,549647683628642
-0,0127194313868060
-0,452682745464152
-0,605877560385565
-1,42706647302949
-35,9107931304111
0,321110515084925 -< this is a position of outlier
36,0098773311071
1,55432187084153
0,329500733819600
-1,10153727341736
-0,535656388090509
0,222372662282162
-0,349206720356847

请先登录，再回答此问题。

采纳的回答

KALYAN ACHARJYA
KALYAN ACHARJYA 2020-12-14
编辑：KALYAN ACHARJYA 2020-12-14
Use diff function, more details
result=diff(data);
You will get a large variation in such cases (Use threshold to get the indecies), hope I understand your question.
  2 个评论
Nik Rocky
Nik Rocky 2020-12-14
Thank you very much! It works!
Here is solution:
res1 = find(diff(v1)>20); %find index of outlier
v1(res1)=rdivide((v1(res1-1))+(v1(res1+1)),2); %enter a mean value from left and right

请先登录，再进行评论。

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Smoothing and Denoising 的更多信息

标签

Community Treasure Hunt

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

Start Hunting!

Translated by