Interpolate NaN on graph

Hello,

I would like to interpolate the missing data on the graph marked by NaN.

when I try it it does not work.

the code is:

filteredData = data(:,2);

vector = [];

for index = 1:length(filteredData)

if filteredData(index) >= 2048

filteredData(index) = NaN;

vector = [vector index];

end

end

index2 = 2;

while index2 < length(filteredData)

if filteredData(index2) < filteredData(index2-1) - 750 && filteredData(index2) < filteredData(index2+1) - 750

vector = [vector index2];

filteredData(index2) = NaN;

index2 = index2 + 1;

end

if filteredData(index2) > filteredData(index2 - 1) + 750 && filteredData(index2) > filteredData(index2+1) + 750

vector = [vector index2];

filteredData(index2) = NaN;

index2 = index2 + 1;

end

index2 = index2 + 1;

end

vector = sort(vector);

y = interp1(1:length(filteredData),filteredData,vector, "linear");

plot(vector, y, '^r');

format long g;

Answer 1

[Star Strider]

If the data have NaN as the value of the dependent variable, and the independent variable is continuous (with no NaN values), then save the original independent variable as a vector, remove the entire row with NaN values from the data, and use the intact (original) independent variable to do the interpolation. The output of interp1 will be the interpolated value of the dependent variable, matching the values of the original independent variable.

[Alex Dimko]

only 3 values are NaN and they are not in the same row.

Plus how do I remove them if we are suppose to assign the values to NaN and then interpolate.

[Star Strider]

I assume that they are dependent variable values, and that all the independent variable values are not NaN.

Remove them by removing all the rows with NaN vlues in either of the dependent variable columns. Then use the original independent variable column (that I assume to be complete and continuous) as the interpolation vector. The result should be the two dependent variable columns with the values interpolated.

Example —

Data = [1 rand(1,2); 2 rand(1,2); 3 rand(1,2); 4 NaN rand; 5 rand(1,2); 6 rand NaN; 7 rand(1,2); 8 rand(1,2)]
nanrows = any(isnan(Data),2);
Dataq = Data(~nanrows,:);
NewData = [Data(:,1) interp1(Dataq(:,1),Dataq(:,[2 3]), Data(:,1))]

.

[Alex Dimko]

how do I delete NaN?

filteredData(index) = []?

[Alex Dimko]

what does Data(~nanrows,:); do? I dont seem to understand the code very well.

doesn't nanrows return true?

[Alex Dimko]

by the way filteredData is only one column of data as only the second column was needed and not all 4.

[Alex Dimko]

YOU don't need to answer any of the questions. I messed around with the code didbsome searching for the meaning online and copied your code. Modified it for one column and it WORKS. I could not believe it. Thanks again. I will probably encounter more problems on this assignment so I will give you more questions. Thanks!

[Star Strider]

‘how do I delete NaN?’

The NaN values are deleted in this assignment:

Dataq = Data(~nanrows,:);

The ‘nanrows’ variable is a logical vector that is true (or 1) where the row has a NaN value and is 0 otherwise. The tilde operator (~) negates that, so this assigns to ‘Dataq’ all the rows that do not have NaN values. See the documentation section on Matrix Indexing for a full explanation.

‘by the way filteredData is only one column of data as only the second column was needed and not all 4.’

If you want to interpolate only the second column, just choose it and the first column (the independent variable) as the ‘Data’ matrix, and use the rest of my code as provided.

Answer 2

[KSSV]

Read about fillmissing.

https://in.mathworks.com/help/matlab/ref/fillmissing.html

[Alex Dimko]

I am requiredto use interp1 by the assignment

[KSSV]

Then you have to use like below:

n = 100 ; 
x = 1:n ;
t = rand(size(x)) ; 

% make some values NaN to fill 
y = t ; 
idx = sort(randperm(n,20)) ;
y(idx) = NaN ; 
% fill nan using interp1 
xi = setdiff(x,idx) ;
yi = y(~isnan(y)) ; 

y(idx) = interp1(xi,yi,idx) ; 

plot(x,t,'r',x,y,'b') 

Actually using random data for demo is not good. But you can follow the procedure.