In paulemms/datamining: Data mining

Introduction

This is Lab 6 on Data Mining.

First we load the datamining package and the data for the lab.

library(datamining)

The data set

Interspersed throughout this lab are questions that you will have to answer at check-off. 1. Download the files for this lab from the course web page to the desktop: http: //www.stat.cmu.edu/~minka/courses/36-350/lab/ 2. Open a Word or Notepad document to record your work. Start R 3. Start -> All Programs -> Class software -> R 1.7.0 4. Load the special functions for this lab: File -—-> Source R code... Browse to the desktop and pick lab6.r (it may have been renamed to lab6.r.txt when you downloaded it). Another window will immediately pop up for you to pick the mining.zip file you downloaded. The dataset o. Lhe dataset 1s the same as lab 0. Load it via data(States) This defines a matrix called States, and a standardized matrix called StatesT. You should use the standardized one. There is also a vector of labels called state.region. Clustering 6. Using the commands from lab 4, use Ward’s method to partition the data into 4 parts. How does the sum of squares of your clustering compare to that of state.region? (. If f isa vector of cluster numbers, this command will name your clusters according state .region: f = factor(f,labels=name.clusters(f,state.region) ) In R, a vector of labels is called a factor. (state.region is also a factor). Projections 8. Using the commands below and trom lab 5, plot the m-projection of the clusters, using state labels and colors. Keep a copy for the homework. 9. Make a dichotomy between South and Not South, and plot the m-projection. Keep a copy for the homework. 10. Repeat for Northeast. Parallel-coordinates 11. Make (and save) a parallel-coordinate plot of the cluster means. 12. Make (and save) two more parallel-coordinate plots: one where Texas is in a group by itself and another where New Mexico is in a group by itself. Note that the variable ordering and scaling may change slightly between the plots. 13. You can now get checked off. Informative projection If x is a matrix and f is a factor, the combination weights for an m-, v-, or mv-projection can be computed via one of: w = projection(x,f,2,type="m") w = projection(x,f,2,type="v") w = projection(x,f,2,type="mv") w can be used just like PCA weights. Scatterplot with colors ‘To make a scatterplot of colored dots or labels, use one of the following: color.plot(v2 ~ vi, x, f) color.plot(v2 ~ vi, x, f, labels=T, asp=1) The first uses dots, the second uses labels. Here x is a matrix and f is a vector of names or numbers. The usual options are also available (like asp). Constructing factors If f is a factor, new factors can be constructed as follows: fd = factor.dichotomy (f,"South") fi = factor.isolate(f,"Texas") fd is a factor with the labels changed to South and NOT South (only two groups). fi is a factor with the label for Texas changed to Texas (thus separating it from the other groups). Parallel-coordinate plot Normally you want to plot the mean of each cluster, rather than all of the data. If x is a matrix and f is a factor: xp = prototypes (x,f) parallel .plot (xp)

paulemms/datamining documentation built on March 1, 2023, 4:01 p.m.