caret ml datasets

There are a number of data sets included in the caret library, but many other examples play around traditional R data sets such as the iris data. A vast resource of machine learning data sets is the UCI ML archive which covers over 350 different datasets. Generally it is important to know first if the data set is geared towards classifications or regressions. Secondly input (x) and output (y) variables need to be known, plus of they are categorial (A,B,C) or continuous (1,2,3) variables. Furthermore it is important to know what the dimension/size is to avoid extremely long examples in the beginning. This is not intended as a comprehensive collection, but rather a quick pointer for practitioners.

---

Example data sets and their dimensions and use View/Download

I recommend the above View/Download button to see the *.csv datasets, the markdown table below is sub-par. Anyway here it goes, check the dimensions to select fast/slow datasets.

Num	Data set	library	Reg/Class	Y-target	X-input	Dimension	Design
1	data(iris)	R	Class	Species	remaining data	150 x 5	multi class
2	data(trees)	R	NA	NA	remaining data	31x 3	multi class
3	data(Glass)	R	Class	Type	remaining data	214 x 10	multi class
4	data(cox2)	caret	Class	cox2Class	cox2Descr,cox2IC50	462x 255	Two class
5	data(oil)	caret	Class	oilType	fattyAcids	96 x 7	multi class
6	data(dhfr)	caret	Reg	Y	remaining data	325 x 229	multi param
7	data(GermanCredit)	caret	Reg/Class	Class	remaining data	1000 x 62	multi param
8	data(BostonHousing)	mlbench	Reg	medv	remaining data	506x14	multi param
9	data(BloodBrain)	caret	Reg	logBBB	bbbDescr	208 x 134	multi param
10	data(mdrr)	caret	Class	mdrrClass	mdrrDescr	528 x 342	Two class
11	data(Satellite)	mlbench	Class	classes	remaining data	6435 x 37	multi class
12	data(cars)	caret	Reg/Class	any	any	804x15	mixed
13	data(dhfr)	caret	Class	Y	remaining data	325x229	Two class
14	data(pottery)	caret	Class	potteryClass	pottery	NA	Two class
15	data(segmentationData)	caret	Class	Class	remaining data	2019 x 61	Two class
16	data(tecator)	caret	Reg	endpoints	absorp	215x 100	multi param
17	data(abalone)	APM	Class	Type	remaining data	4177 x 9	multi class
18	data(AlzheimerDisease)	APM	Class	diagnosis	predictors	333 x 130	Two class
19	data(ChemicalManufacturingProcess)	APM	Reg	Yield	remaining data	176 x 58	multi param
20	data(concrete)	APM	Reg	CompressiveStrength	remaining data	1030 x 9	multi param
21	data(FuelEconomy)	APM	Reg/Class			3 sets	multi param
22	data(hepatic)	APM	Class	injury	remaining data	2 sets	multi param
23	data(solubility)	APM	Reg	solTestY		1267 x 228	multi param
24	data(permeability)	APM	Reg	permeability	fingerprints	165 x 1107	multi param
25	data(schedulingData)	APM	Class	Class	remaining data	4331x8	multi class
26	data(segmentationOriginal)	APM	Class	Class	remaining data	2019 x 119
27	data(twoClassData)	APM	Class	classes	predictors	208 x 2	Two class
28	data(BreastCancer)	mlbench	Class	Class	remaining data	699 x 11	Two class
29	data(PimaIndiansDiabetes)	mlbench	Class	diabetes	remaining data	768 x 9	Two class
30	data(Sonar)	mlbench	Class	Class	remaining data	208x61	Two class
31	Human Activity Recognition (HAR)	puc-rio.br	Class	“class”	remaining data	165634 x 21	multi class

---

Show all available data sets loaded

# show all available data sets installed
library(caret); library(datasets); library(AppliedPredictiveModeling); library(mlbench); data();

# Data sets in package ‘AppliedPredictiveModeling’:
# ChemicalManufacturingProcess                          Chemical Manufacturing Process Data
# abalone                                               Abalone Data
# bio (hepatic)                                         Hepatic Injury Data
# cars2012 (FuelEconomy)                                Fuel Economy Data
# chem (hepatic)                                        Hepatic Injury Data
# classes (twoClassData)                                Two Class Example Data
# concrete                                              Compressive Strength of Concrete from Yeh (1998)
# ...

---

Quick instructions to work with the data sets

# load the dataset
data(iris)
# get the dimension of the dataset
dim(iris)
## [1] 150   5
# get the class name (here data frame)  to choose correct operators
class(iris)
## [1] "data.frame"

# invoke simple data viewer
View(iris)
# invoke the useless editor
edit(iris)
# get the data structure
str(iris)
> str(iris)
##'data.frame':   150 obs. of  5 variables:
## $ Sepal.Length: num  5.1 4.9 4.7 4.6 5 5.4 4.6 5 4.4 4.9 ...
## $ Sepal.Width : num  3.5 3 3.2 3.1 3.6 3.9 3.4 3.4 2.9 3.1 ...
## $ Petal.Length: num  1.4 1.4 1.3 1.5 1.4 1.7 1.4 1.5 1.4 1.5 ...
## $ Petal.Width : num  0.2 0.2 0.2 0.2 0.2 0.4 0.3 0.2 0.2 0.1 ...
## $ Species     : Factor w/ 3 levels "setosa","versicolor",..: 1 1 

# load the package that contains the data
require(caret)
# load the mdrr regression data set 
data(mdrr)
# view y-factor (View(mdrr) does not work)
View(mdrrClass)
# get dimension of mdrrClass
length(mdrrClass)
## [1] 528
dim(mdrrDescr)
## [1] 528 342

---

Code

• ML data sets - view some important ML datasets and their dimensions.

LINKS

• mlbench data sets - R mlbench package which covers many data sets
• caret data sets - R caret PDF contains detailed info about the caret specific sets
• caret WIKI - quick description of caret sets
• ApplPredMod data sets - from the compendium book Applied Predictive Modeling
• UCI sets - general collection (many already converted to R)