Debian Codegraph @

Testing/Data/Input/Statistics/TwoDimensionTwoGaussian.Readme

raw
1 * Overview
2 The "TwoDimensionTwoGaussian.dat" file has 2000 measurement
3 vectors. Each measurement vector consist of two components of
4 double. The measurement vectors are generated using GNU R
5 statistical package (https://www.r-project.org).
6
7 Half of the measurement vectors are generated using the the
8 "First Gaussian" class statistics and the other half are
9 generated using the "Second Gaussian" class statistics in the
10 following section. First 1000 rows in the file belong to the
11 first class and the rest belong to the second class.
12
13 The data is generated by a multivariate random normal variate
14 generator (rmvnorm function in mvtnorm package). The exact GNU
15 R script for genation is in the "GNU R script" section of this
16 document.
17
18 * Class Statistics
19 1) class statistics for random sample generation
20 First Gaussian class:
21 mean
22 100 100
23 covariance
24 800 0
25 0 800
26
27 Second Gaussian class:
28 mean
29 200 200
30 covariance
31 800 0
32 0 800
33 2) class statistics of the generated samples
34 First Gaussian class:
35 mean
36 99.261 100.078
37 covariance
38 814.95741 38.40308
39 38.40308 817.64446
40
41
42 Second Gaussian class:
43 mean
44 200.1 201.3
45 covariance
46 859.785295 -3.617316
47 -3.617316 848.991508
48
49 * GNU R Script
50
51 library(mvtnorm)
52 mean1 = c(100, 100)
53 mean2 = c(200, 200)
54 sigma = matrix(data=c(800, 0, 0, 800), ncol=2, nrow=2, byrow=T)
55 gauss1 = rmvnorm(1000, mean1, sigma)
56 gauss2 = rmvnorm(1000, mean2, sigma)
57 mydata = rbind(gauss1, gauss2)
58 write.table(mydata, file="TwoDimensionTwoGaussian.dat", quote=F, row.names=F, col.names=F)