Modified Spline Regression

This is the repository for our paper titled "Modified Spline Regression for Neual Spiking Data" available on PLOS ONE journal. Here we provide our MATLAB code for a simulated neural spiking data.

Code

The code is divided to four categoreis:

1. The main.m MATLAB file contains the main framework to run the experiment.
2. The SimulateSpike.m file generates simulated neural spiking data
3. The Hist.m file is to build design matrix for multiplicative hisory model
4. The Indicator.m, RaisedCos.m, CardinalSpline.m and ModifiedCardinalSpline.m are four basis functions that used for point process generalizes linear model (GLM).

To run the model, you should clone all the files and run main.m file.

Results

The following figure shows point process GLM fit using four different basis function on the simulated data.

In order to quantify the effect of the choice of basis function on the size of the confidence bounds, we computed a confidence interval width ratio (CIWR) of the confidence interval width at the end points over the average of confidence interval width in the interior regions for four different bases.

First Header	Indicator	Raised Cosine	Cardinsl Spline	Modified Cardinal Spline
CIWR_start	0.0047	1.183	20.152	1.119
CIWR_end	0.0104	0.641	2.682	1.264