Asked by Duphrin
on 21 Oct 2019

I have an excel file which containes more than 300000 in both column a and b. If the values in first column is less than 1 ,I need to average the corrosponding values in the second column and then increasing the range to 1 to 2 and so on up to 30..Any help would be greatly appreciated. Thank you.

My file looks like this

2.2060 0.3120

2.2140 0.3138

2.2180 0.3146

2.2260 0.3164

3.8920 0.2611

3.9000 0.259

3.9070 0.2571

3.8840 0.2632

3.9600 0.2431

4.0010 0.2322

If the values in first column is less than 1 ,I need to average the corrosponding values in the second column and then increasing the range to 1 to 2 and so on up to 30..Any help would be greatly appreciated. Thank you.

Answer by Joe Vinciguerra
on 21 Oct 2019

Accepted Answer

Here's another approach. Because of the size of your dataset I would recommend avoiding loops. the accumarray function applies similar logic as suggested by Nicolas, but can run faster.

yourData(:,3) = floor(yourData(:,1)); % create a new column by rounding column 1 down to nearest integer

% ...OR...

yourData(:,3) = discretize(yourData(:,1),0:30,'IncludedEdge','left'); % this is SIMILAR floor, but gives you more control in defining how to handle the edges of your data if you need to.

% choose only one of the above based on your needs.

[C,~,IC] = unique(yourData(:,3)); % C is a list of all unique values in the new 3rd column (integers between 0 and 30). IC associates the row numbers in your data to the row numbers of C.

avg = accumarray(IC, yourData(:, 2), [], @mean); % calculate the mean in your data of column 2 for rows in IC with identical elements

Result = [C avg];

Sign in to comment.

Answer by Nicolas B.
on 21 Oct 2019

For that situation, the easiest way is to select data based on a criteria. For example, if your table is named t:

indRows = t(:,1) < 1; % get row indexes for values below 1

myMean = mean(t(indRows, 2)); % compute the mean

Of course, you can also merge the 2 line of codes to avoid using indRows.

Duphrin
on 21 Oct 2019

Nicolas B.
on 21 Oct 2019

Then, I think that would be the easiest solution to simply loop with your criterias. I could imagine solutions based on the function discretize(), but it wouldn't avoid a loop:

mylim = [-inf, 1:30, inf]; % list of your limits

y = discretize(t, mylim); % you get the indexes

m = NaN(size(myLim)); % your output vector

for n = 1:numel(m) % loop to get the mean value

m(n) = mean(t(y == n), 'omitnan');

end

Sign in to comment.

Opportunities for recent engineering grads.

Apply Today
## 0 Comments

Sign in to comment.