if your database is M , that is

M = [
0.00 0 -1 -1 -127 -1 -1 52 -1
0.25 0 -1 -1 -127 -1 -1 52 -1
0.50 0 -1 -1 -127 -1 -1 51 -1
0.75 0 -1 -1 -127 -1 -1 51 -1
....
4.50 121 -1 -1 -127 -1 -1 56 -1
4.75 120 -1 -1 -127 -1 -1 52 -1
];

is to use a simple and rude for loop:

MSub = M(:,2); % 
skipCountCache = 0; % 0
finalSkipCount = 0; % 
accumsum = 0; %
for i = 1:numel(MSub)
    if M(i) == 0
        skipCountCache = skipCountCache + 1;
        if skipCountCache > 80 % 80
            break;
        end
        continue;
    end

    if skipCountCache > 20
        skipCountCache = 0; % 20
    else                
        finalSkipCount = finalSkipCount + skipCountCache; % finalSkipCount
        skipCountCache = 0; % 
    end
            
    accumsum = accumsum + MSub(i);
end

finalSkipCount = finalSkipCount + skipCountCache; 0finalSkipCount
average = accumsum / (i - finalSkipCount);

for loop is easy to understand and write, but not efficient, especially if you have a large amount of data.

at this point, we can create a filter consisting of 0 and 1 (essentially a href=" https://uk.mathworks.com/help/matlab/ref/logical.html" rel= "nofollow noreferrer" > logical array ). Then use find and diff to process the filter.

MSub = M(:,2);
zeroFilter = MSub == 0; % 
boundaryFinder = diff(zeroFilter); % ;0110
openingZeroIndices = find(boundaryFinder == 1); % 00index-1
closingZeroIndices = find(boundaryFinder == -1); % 00index
endWithZero = false; 
if MSub(1) == 0 % 0
    openingZeroIndices = [0; openingZeroIndices];
end
if MSub(end) == 0 % 0
    endWithZero = true;
    closingZeroIndices = [closingZeroIndices; numel(MSub)];
end

assert(numel(openingZeroIndices) == numel(closingZeroIndices)); % openingZeroclosingZero

zeroGroupSizes = closingZeroIndices - openingZeroIndices;  % 0
assert(all(zeroGroupSizes>0)); % 0

terminateIndex = openingZeroIndices(zeroGroupSizes >= 80); % 080

if isempty(terminateIndex) % 080
    if endWithZero
        terminateIndex = openingZeroIndices(end);
    else
        terminateIndex = numel(MSub);
    end
end

finalSkipCount = sum(zeroGroupSizes(zeroGroupSizes <= 20)); % 
average = sum(MSub(1:terminateIndex)) / (terminateIndex - finalSkipCount); % 

this is really beyond expectations. Thank you!