From choosing the numbers of factors to extract to inspecting loadings, factor analysis can be very visual in nature. The FAtools R package aims to make this process easier by providing functions to do visualizations with ease.
library('devtools')
#devtools::install_github('mattkcole/FAtools')
library('FAtools')We can first look at our data (here we are using the possibly cliche but familiar data, mtcars).
library(datasets)
summary(mtcars)
#> mpg cyl disp hp
#> Min. :10.40 Min. :4.000 Min. : 71.1 Min. : 52.0
#> 1st Qu.:15.43 1st Qu.:4.000 1st Qu.:120.8 1st Qu.: 96.5
#> Median :19.20 Median :6.000 Median :196.3 Median :123.0
#> Mean :20.09 Mean :6.188 Mean :230.7 Mean :146.7
#> 3rd Qu.:22.80 3rd Qu.:8.000 3rd Qu.:326.0 3rd Qu.:180.0
#> Max. :33.90 Max. :8.000 Max. :472.0 Max. :335.0
#> drat wt qsec vs
#> Min. :2.760 Min. :1.513 Min. :14.50 Min. :0.0000
#> 1st Qu.:3.080 1st Qu.:2.581 1st Qu.:16.89 1st Qu.:0.0000
#> Median :3.695 Median :3.325 Median :17.71 Median :0.0000
#> Mean :3.597 Mean :3.217 Mean :17.85 Mean :0.4375
#> 3rd Qu.:3.920 3rd Qu.:3.610 3rd Qu.:18.90 3rd Qu.:1.0000
#> Max. :4.930 Max. :5.424 Max. :22.90 Max. :1.0000
#> am gear carb
#> Min. :0.0000 Min. :3.000 Min. :1.000
#> 1st Qu.:0.0000 1st Qu.:3.000 1st Qu.:2.000
#> Median :0.0000 Median :4.000 Median :2.000
#> Mean :0.4062 Mean :3.688 Mean :2.812
#> 3rd Qu.:1.0000 3rd Qu.:4.000 3rd Qu.:4.000
#> Max. :1.0000 Max. :5.000 Max. :8.000Let's first make our correlation matrix - we wont worry about scaling or investigating our data much for this demonstration (usually a bad idea).
corr.matrix <- cor(mtcars)Let's load the packages we need for our analysis:
library('psych') # for statistical methods
library('FAtools') # for some plotting and EDA
library('dplyr') # for data wrangling
library('knitr') # for rmd helpLets make and plot our scree plot to assess the number of factors present.
s.plot <- FAtools::scree_plot(corr.matrix, nrow(mtcars), ncol(mtcars))
plot(s.plot)
We can conduct our factor analysis with two factors using the psych package.
results <- psych::fa(corr.matrix, 2, rotate = "varimax")
results$loadings
#>
#> Loadings:
#> MR1 MR2
#> mpg 0.675 -0.630
#> cyl -0.634 0.731
#> disp -0.727 0.607
#> hp -0.316 0.881
#> drat 0.812 -0.219
#> wt -0.784 0.454
#> qsec -0.151 -0.873
#> vs 0.295 -0.788
#> am 0.901
#> gear 0.882 0.150
#> carb 0.809
#>
#> MR1 MR2
#> SS loadings 4.464 4.393
#> Proportion Var 0.406 0.399
#> Cumulative Var 0.406 0.805The loadings look pretty good, but we can make them more interpretable by excluding low loadings (param: cutoff), rounding (param: roundto), incorporate a data dictionary, and include labels -- And we can use the knitr::kable() function for great looking tables in Rmarkdown documents.
FAtools::loadings_table(results$loadings, 2, cutoff = 0.3, roundto = 2) %>%
kable()| name | V1 | V2 |
|---|---|---|
| mpg | 0.68 | -0.63 |
| cyl | -0.63 | 0.73 |
| disp | -0.73 | 0.61 |
| hp | -0.32 | 0.88 |
| drat | 0.81 | |
| wt | -0.78 | 0.45 |
| qsec | -0.87 | |
| vs | 0.3 | -0.79 |
| am | 0.9 | |
| gear | 0.88 | |
| carb | 0.81 |
Say we had more informative names than colnames(mtcars).
cool_names <- c("Miles Per Gallon", "Cylinders", "Displacement",
"Gross horsepower", "Rear Axle ratio", "Weight (1K lbs)",
"1/4 mile time", "V/S", "Manual", "Forward gears",
"Carburetors")And say we wern't really all that interested in loadings with an absolute value less than 0.3.
FAtools::loadings_table(loading_frame = results$loadings,
cutoff = 0.3, roundto = 2,
Name = colnames(mtcars),
description = cool_names) %>%
kable()| name | description | V1 | V2 |
|---|---|---|---|
| mpg | Miles Per Gallon | 0.68 | -0.63 |
| cyl | Cylinders | -0.63 | 0.73 |
| disp | Displacement | -0.73 | 0.61 |
| hp | Gross horsepower | -0.32 | 0.88 |
| drat | Rear Axle ratio | 0.81 | |
| wt | Weight (1K lbs) | -0.78 | 0.45 |
| qsec | 1/4 mile time | -0.87 | |
| vs | V/S | 0.3 | -0.79 |
| am | Manual | 0.9 | |
| gear | Forward gears | 0.88 | |
| carb | Carburetors | 0.81 |
We could also display this graphically, which works well when we have more retained factors or many more variables. (let's say we have 5 factors extracted).
loadings5 <- cor(mtcars) %>%
psych::fa(2, rotate = "varimax")
FAtools::loadings_plot(loadings5$loadings,
colorbreaks = c(0, 0.2, 0.4, 0.6, 0.8, 1),
labRow = c("F1", "F2"),
columnlabels = cool_names)
#> Warning in if (is.na(labRow) == T) {: the condition has length > 1 and only
#> the first element will be used
Looks great!
Submit and issue with any concerns!
Credits: Much of the scree plot functionality comes from code provided by: www.statmethods.net
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent