The algorithm discussed in class is in the PPT
In this assignment you will implement the reconstruction algorithm discussed in class. You will also perform a comprehensive assessment of the performance of the algorithm and generate multiple plots.
Performance assessment
You should test your implementation on at least five different distributions for the original data. Use a Gaussian distribution for the noise data. Your experiments should explore the following aspects of the algorithm:
reconstruction quality when data comes from different distributions (you must test at least five)
impact of the amount of noise (set using the mean and variance of the Gaussian noise function) on the reconstruction quality
impact of data size (number of samples you have)
impact of interval length
impact of the number of iterations
For each experiment, make a plot similar to the one below. The original data (20000 samples here) in this experiment came from a mixture of two Gaussian distributions. The noise came from a Gaussian distribution with zero mean and unit variance. The interval size is 0.2. The R code for the data generation is as follows:
# Original data
x <- c(rnorm(10000,mean=-1,sd=0.5),rnorm(10000,mean=1,sd=0.25))
# Noise data
y <- rnorm(20000,mean=0,sd=1)
# Perturned data
w <- x+y
Report
Include parameters for each experiment and the resulting plots in a pdf file.
Attachment:- Chapter.zip