Finding the first index in the datasets with a gap

Hi there,

I have some data, which is attached, and I'd like to try to find the first index in the datasets. So I write some code, which I've attached, but I'm having trouble finding some indexes.

For example, there is a gap between finding the number 4 and finding the first index of 4 after this gap (figure). I also like to repeat the process for other numbers.

Any suggestions would be greatly appreciated.

Answer 1

[Voss]

Let me reproduce your image first:

% load and plot data
data = readmatrix('Data.xls');
plot(data(:,1),data(:,2),'r','LineWidth',2)
hold on

% get the set of unique values in data(:,2), which is 1-7 in this case
vals = unique(data(~isnan(data(:,2)),2))

vals = 7×1

1 2 3 4 5 6 7

% find the first index of each value
Nvals = numel(vals);
first_idx = zeros(Nvals,1);
for ii = 1:Nvals
    first_idx(ii) = find(data(:,2) == vals(ii),1);
end

% plot a blue marker at the first index of each value
plot(data(first_idx,1),data(first_idx,2),'ob','MarkerFaceColor','b')

Now, I'll include some code for finding the first index after a gap:

% initialize post-gap indices to NaN (not all values have a gap - use NaN to tell which)
post_gap_idx = NaN(Nvals,1);

for ii = 1:Nvals
    
    % Find the first location after the first instance of vals(ii) in data(:,2)
    % where data(:,2) is NOT equal to vals(ii). That is, after data(:,2) goes
    % to vals(ii), non_val_idx is the first location where data(:,2) has some 
    % other value:
    non_val_idx = find(data(first_idx(ii)+1:end,2) ~= vals(ii),1) + first_idx(ii);
    
    if isempty(non_val_idx)
        % empty non_val_idx means that once data(:,2) goes to vals(ii), it
        % remains there, so no gap exists and there's nothing to do in that case
        continue
    end
    
    % here, we know that data(:,2) goes to vals(ii) and later changes to some 
    % other value, so now find the location where data(:,2) changes back to vals(ii):
    back_to_val_idx = find(data(non_val_idx+1:end,2) == vals(ii),1) + non_val_idx;
    
    if isempty(back_to_val_idx)
        % empty back_to_val_idx means that data(:,2) never returns to vals(ii)
        % once it switches away, so in this case as well there's no gap
        continue
    end
    
    % here, we know that data(:,2) changes back to vals(ii) after having switched 
    % away (i.e., a gap), so store the location at which this happens as 
    % post_gap_idx(ii):
    post_gap_idx(ii) = back_to_val_idx;
    
end

post_gap_idx

post_gap_idx = 7×1

43091 34961 NaN 22678 NaN NaN NaN

% only use the non-NaN values for plotting
temp = post_gap_idx(~isnan(post_gap_idx))

temp = 3×1

43091 34961 22678

% plot a green marker at the post-gap locations
plot(data(temp,1),data(temp,2),'gs','MarkerFaceColor','g')

[payam samadi]

well said, thank you very much.

[Voss]

You're welcome!