smoothing datasets containg nan

how i can smooth these data :

data=[1 1.927259

2 1.519675

3 2.553609

4 3.138377

5 3.6647

6 6.6849

7 6.937631

8 6.087666

9 nan

10 9.182496

11 9.536923

12 8.610497

13 8.3172

14 17.25756

15 6.725476

16 18.0788

17 8.0282

18 9.9958

19 7.063844

20 5.164532

21 3.5716

22 2.296498

23 1.757754

24 1.559203];

Answer 1

[Les Beckham]

Interestingly, the default window size for the 'sgolay' smoothing method in the smoothdata function generates smoothed data that is indistinguishable from the original data in this case.

Below, I've made a slight modification to the answer I gave to your previous question (found here):

https://www.mathworks.com/matlabcentral/answers/2038326-how-to-smooth-data?s_tid=srchtitle

I added the window size optional argument to the smoothdata call below. Adjust as desired. Or, experiment with other smoothing methods. See the link above to the documentation.

data=[ ...
1  1.927259  
2  1.519675  
3  2.553609   
4  3.138377   
5  3.6647    
6  6.6849   
7  6.937631 
8  6.087666    
9  nan 
10  9.182496  
11  9.536923   
12  8.610497  
13  8.3172   
14  17.25756
15  6.725476 
16  18.0788  
17  8.0282  
18  9.9958   
19  7.063844 
20  5.164532  
21  3.5716   
22  2.296498 
23  1.757754
24  1.559203];
plot(data(:,1), data(:,2))
hold on
window_size = 8; % <<< experiment with adjusting this to get results that you like
sm_data = smoothdata(data(:,2), 'sgolay', window_size); % save the smoothed data
plot(data(:,1), sm_data)
grid on
legend('original data', 'smoothed data')

Also, please don't close questions minutes after posting them. Give people a chance to answer them.

[ahmad Saad]

@Les Beckham

Deep thanks for attention and kind help.

I apologize. Not me, probably MathWorks does.

[ahmad Saad]

@Les Beckham

are there other methods that can do perfect smoothing

[Les Beckham]

@ahmad Saad

You are quite welcome. Don't forget to Accept the answer if it solved your issue.

What would you consider "perfect smoothing"?

There are lots of methods of smoothing data in Matlab (some of which require additional toolboxes).

Just use your favorite search engine or the Matlab documentation search tool to explore your options.

[ahmad Saad]

@Les Beckham

thanks again

i consider smoothing means not to have broken lines

[Les Beckham]

If filling in gaps in your data is your primary goal (rather than smoothing out jagged parts of the curve), I would recommend reading the documentation for fillmissing. That function can also do some smoothing of the jagged parts, like smoothdata.

Here is an example using this set of data, just filling in the gaps using linear interpolation.

Note that I swapped the order of the plots (plotting the original data on top of the smoothed data to make it more obvious where the gap was).

data=[ ...
1  1.927259  
2  1.519675  
3  2.553609   
4  3.138377   
5  3.6647    
6  6.6849   
7  6.937631 
8  6.087666    
9  nan 
10  9.182496  
11  9.536923   
12  8.610497  
13  8.3172   
14  17.25756
15  6.725476 
16  18.0788  
17  8.0282  
18  9.9958   
19  7.063844 
20  5.164532  
21  3.5716   
22  2.296498 
23  1.757754
24  1.559203];
sm_data = fillmissing(data(:,2), 'linear'); % save the smoothed data
plot(data(:,1), sm_data)
hold on
plot(data(:,1), data(:,2), 'o-')
grid on
legend('smoothed data', 'original data')

[ahmad Saad]

@Les Beckham

thanks again for your kind attention.

Actually, i need to smooth the data not to fill the gaps.

[Image Analyst]

Well the obvious thing to do is to fill the gaps (which you said once that you needed to do), and then after that smooth the data.

Answer 2

[Image Analyst]

Have you seen the "DataCleaner" app on the Apps tab of the tool ribbon? Try it.

[ahmad Saad]

@Image Analyst

Thanks for response.

Unfortunally, I havent this app

[Image Analyst]

Yes you do. It's included in base MATLAB. Did you go to the Apps tab? Did you click on the little down arrow on the very right side of the group of all the applet icons? Look in the group/row called "MATLAB". See image below.