This problem is made much simpler if you leverage the capabilities of cdfread. It can read in cdfepochs as datetimes, simplifiing that step. You can then use movmean to create a 3-hour moving mean (see this example).
% Obtain info about what is in the CDF file
filename = "SW_EXTD_EFIA_LP_HM_20131216T000000_20131216T235959_0106.cdf";
ci = cdfinfo(filename)
ci.Variables
% Load the specified variables
data = cdfread(filename,'Variables',{'n','Timestamp'},"DatetimeType","datetime"); % Electron density, Time
Ne = vertcat(data{:,1}); % Electron density
ti = vertcat(data{:,2}); % Time
% compute a 3 hour moving mean
avg_Ne = movmean(Ne, hours(3), 'SamplePoints', ti)
% Visualize
plot(ti, avg_Ne);
xlabel('Time');
ylabel('Ne');
title('3-Hour Time Average');
Since the data is too large to load here, here is a screenshot of what the final plot looked like for me. I have not gone through the data to verify any of the results.
