Local Estimation a spike firing rate (in spikes per unit time). The estimation procedure approximates the log of the spike firing rate by a quadratic polynomial, within a sliding window. The first argument to locfit() is a column vector of the spike times. 'family','rate' specifies that the output of the density estimate should be in terms of events per unit time. (instead of the default, statistical density estimation). 'alpha',0.6 specifies the width of the sliding windows, as a fraction of the total spikes. xlim gives the limits of the observation interval. Correct specification of this is critical to avoid introducing bias at end-points. The lfband(fit) line adds confidence bands (based on pointwise 95% coverage) to the plot. The spike time data is from Hemant Bokil.
0001 % Local Estimation a spike firing rate (in spikes per unit time). 0002 % 0003 % The estimation procedure approximates the log of the spike firing 0004 % rate by a quadratic polynomial, within a sliding window. 0005 % 0006 % The first argument to locfit() is a column vector of the spike times. 0007 % 'family','rate' specifies that the output of the density estimate should 0008 % be in terms of events per unit time. (instead of the default, statistical 0009 % density estimation). 0010 % 0011 % 'alpha',0.6 specifies the width of the sliding windows, as a fraction of 0012 % the total spikes. 0013 % 0014 % xlim gives the limits of the observation interval. Correct specification 0015 % of this is critical to avoid introducing bias at end-points. 0016 % 0017 % The lfband(fit) line adds confidence bands (based on pointwise 95% 0018 % coverage) to the plot. 0019 % 0020 % The spike time data is from Hemant Bokil. 0021 0022 load lmem5312.mat; 0023 fit = locfit(data(6).times{1},'xlim',[78.9 80.30],'family','rate','nn',0.6); 0024 lfplot(fit); 0025 title('Spike Firing Rate Estimation'); 0026 lfband(fit);