Download presentation
Presentation is loading. Please wait.
Published byScot Thornton Modified over 9 years ago
2
Social Media Application
4
Goal: Data Reduction for Data Visualization
5
People Variables CLUSTER ANALYSIS FACTOR ANALYSIS Variable/Dimension Reduction Cluster and Factor Analysis
6
For car buying, what matters to customers?
7
Brainstorm: Car Purchase
8
Surveys
9
QRate on a scale of 1-Low to 9-High (randomized list) Shopper#1New BMW 1971 Olds 442 Conv. 1Initial Price934 2Style789 3# of Miles on Car794 4Reliability762 5Color579 6Comfort675 7Horsepower269 8Safety671 9Financing Terms752 10Country Origin177 11Drive Type (Front, 4WD)446 12Miles Per Gallon (MPG)675 13Warranty Coverage452 Survey: Attribute Ratings Many more features, options….
10
QRate on a sale of 1- 9 1Initial Price 2Style 3# of Miles on Car 4Reliability 5Color 6Comfort 7Horsepower 8Safety 9Financing Terms 10Country Origin 11Drive Type (Front, 4WD) 12Miles Per Gallon (MPG) 13Warranty Coverage Survey: Attribute Ratings 12345678910101 1212 1313
11
cor(data, digits=2) Correlation Matrix
13
install.packages("corrgram") library(corrgram) corrgram(data)
14
Factor Analysis / Variable Reduction Correlation Matrix Correlated variables are grouped together and separated from other variables with low or no correlation Factor Analysis
15
F1 Factor Analysis F2FN …. F3
16
First & Second Principal Components Z 1 and Z 2 are two linear combinations. Z 1 has the highest variation (spread of values) Z 2 has the lowest variation 16
17
F1 b’s Factor Loadings Factor Analysis F2FN …. F3
19
Packages LibraryPC MethodRotationPlot psychfa()YesNo principal() princomp()NoYes
20
Principal Components Analysis
21
Model model <- princomp(data, cor=TRUE) summary(model) biplot(model)
22
Output # scree plot plot(fit,type="lines")
23
Psych Package
24
Psych Package – fa library(psych) rmodel <- fa(r = corMat, nfactors = 3, rotate = “none", fm = "pa")
25
Each variable (circle) loads on both factors and there is no clarity about separating the variables into different factors, to give the factors useful names. Factor 2 Factor 1 Rotation Rotations Courtesy of Professor Paul Berger
26
26 “CLASSIC CASE” After rotation of ~45 0 NOW, all variables are loading on one factor and not at all the other; This is an overly “dramatic” case. Not Correlated Orthogonal Varimax = Orthogonal Rotation Rotation Rotations Courtesy of Professor Paul Berger
27
Psych Package – fa library(psych) rmodel <- fa(r = corMat, nfactors = 3, rotate = "oblimin", fm = "pa")
28
Psych Package – principal library(psych) fit <- principal(ratings6, nfactors=4, rotate=“null")
29
Psych Package – principal library(psych) fit <- principal(ratings6, nfactors=4, rotate="varimax“) corrgram(ratings6[,(1,2,9,12,3,4,6,8,10,5,11,7,13)]) Orthogonal / No Correlation
30
3 Factor vs. 4 Factor
31
Style Comfort Color Upgrade Packages Reliability Safety Country Origin Horsepower Nice Dash Miles Per Gallon Initial Price # of Miles on Car Financing Options Aaahh!!! Factor Money
32
Perceptual Map Factor Loadings Brand Ratings Weights Average Variance
33
Which One? Which Car? Price $$$ $ Sweet!!!BORING Aaaah factor…
34
Factor Analysis Recap
35
Dimensionality Reduction Applications Algorithms
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.