fdth (version 1.3-0)

fdth-package: Frequency distribution tables, histograms and polygons

Description

The fdth package contains a set of functions which easily allows the user to make frequency distribution tables (fdt), its associated histograms and frequency polygons (absolute, relative and cumulative). The fdt can be formatted in many ways which may be suited to publication in many different ways (papers, books, etc). The plot method (S3) is the histogram which can be dealt with the easiness and flexibility of a high level function.

Author

Faria, J. C.
Allaman, I. B
Jelihovschi, E. G.

Details

The frequency of a particular observation is the number of times the observation occurs in the data. The distribution of a variable is the pattern of frequencies of the observation.

Frequency distribution table fdt can be used for ordinal, continuous and categorical variables.

The R environment provides a set of functions (generally low level) enabling the user to perform a fdt and the associated graphical representation, the histogram. A fdt plays an important role to summarize data information and is the basis for the estimation of probability density function used in parametrical inference.

However, for novices or ocasional users of R, it can be laborious to find out all necessary functions and graphical parameters to do a normalized and pretty fdt and the associated histogram ready for publications.

That is the aim of this package, i.e, to allow the user easily and flexibly to do both: the fdt and the histogram. The most common input data for univariated is a vector. For multivariated data can be used both: a data.frame, in this case also allowing grouping all numerical variables according to one categorical, or matrices.

The simplest way to run fdt and fdt_cat is by supplying only the x object, for example: d <- fdt(x). In this case all necessary default values (breaks and right) ("Sturges" and FALSE respectively) will be used, if the x object is categorical then just use d <- fdt_cat(x).

If the varable is of contiuos type, you can also supply:

• x and k (number of class intervals);

• x, start (left endpoint of the first class interval) and end (right endpoint of the last class interval); or

• x, start, end and h (class interval width).

These options make the fdt very easy and flexible.

The fdt and fdt_cat object store information to be used by methods summary, print and plot. The result of plot is a histogram or polygon (absolute, relative or cumulative). The methods summary, print and plot provide a reasonable set of parameters to format and plot the fdt object in a pretty (and publishable) way.

See Also

hist provided by graphics and table, cut both provided by base.

Examples

library (fdth)

# Numerical
#======================
# Vectors: univariated
#======================
x <- rnorm(n=1e3,
           mean=5,
           sd=1)

(ft <- fdt(x))

# Histograms
plot(ft)  # Absolute frequency histogram

plot(ft,
     main='My title')

plot(ft,
     x.round=3,
     col='darkgreen')

plot(ft,
     xlas=2)

plot(ft,
     x.round=3,
     xlas=2,
     xlab=NULL)

plot(ft,
     v=TRUE,
     cex=.8,
     x.round=3,
     xlas=2,
     xlab=NULL,
     col=rainbow(11))

plot(ft,
     type='fh')    # Absolute frequency histogram

plot(ft,
     type='rfh')   # Relative frequency histogram

plot(ft,
     type='rfph')  # Relative frequency (%) histogram

plot(ft,
     type='cdh')   # Cumulative density histogram

plot(ft,
     type='cfh')   # Cumulative frequency histogram

plot(ft,
     type='cfph')  # Cumulative frequency (%) histogram

# Polygons
plot(ft,
     type='fp')    # Absolute frequency polygon

plot(ft,
     type='rfp')   # Relative frequency polygon

plot(ft,
     type='rfpp')  # Relative frequency (%) polygon

plot(ft,
     type='cdp')   # Cumulative density polygon

plot(ft,
     type='cfp')   # Cumulative frequency polygon

plot(ft,
     type='cfpp')  # Cumulative frequency (%) polygon

# Density
plot(ft,
     type='d')     # Density

# Summary
ft

summary(ft)  # the same

print(ft)    # the same

show(ft)     # the same

summary(ft,
        format=TRUE)      # It can not be what you want to publications!

summary(ft,
        format=TRUE,
        pattern='%.2f')   # Huumm ..., good, but ... Can it be better?

summary(ft,
        col=c(1:2, 4, 6),
        format=TRUE,
        pattern='%.2f')   # Yes, it can!

range(x)                  # To know x

summary(fdt(x,
            start=1, 
            end=9,
            h=1),
        col=c(1:2, 4, 6),
        format=TRUE,
        pattern='%d')     # Is it nice now?

# The fdt.object
ft[['table']]                        # Stores the feq. dist. table (fdt)
ft[['breaks']]                       # Stores the breaks of fdt
ft[['breaks']]['start']              # Stores the left value of the first class
ft[['breaks']]['end']                # Stores the right value of the last class
ft[['breaks']]['h']                  # Stores the class interval
as.logical(ft[['breaks']]['right'])  # Stores the right option

# Theoretical curve and fdt
y <- rnorm(1e5,
           mean=5, 
           sd=1)

ft <- fdt(y,
          k=100)

plot(ft,
     type='d',                       # density
     col=heat.colors(100))

curve(dnorm(x,
            mean=5, 
            sd=1),
      n=1e3,      
      add=TRUE, 
      lwd=4)

#=============================================
# Data.frames: multivariated with categorical
#=============================================
mdf <- data.frame(X1=rep(LETTERS[1:4], 25),
                  X2=as.factor(rep(1:10, 10)),
                  Y1=c(NA, NA, rnorm(96, 10, 1), NA, NA),
                  Y2=rnorm(100, 60, 4),
                  Y3=rnorm(100, 50, 4),
                  Y4=rnorm(100, 40, 4),
                  stringsAsFactors=TRUE)

#(ft <- fdt(mdf)) # Error message due to presence of NA values

(ft <- fdt(mdf,
           na.rm=TRUE))

# Histograms
plot(ft,
     v=TRUE)

plot(ft,
     col=rainbow(8))

plot(ft,
     type='fh')

plot(ft,
     type='rfh')

plot(ft,
     type='rfph')

plot(ft,
     type='cdh')

plot(ft,
     type='cfh')

plot(ft,
     type='cfph')

# Poligons
plot(ft,
     v=TRUE,
     type='fp')

plot(ft,
     type='rfp')

plot(ft,
     type='rfpp')

plot(ft,
     type='cdp')

plot(ft,
     type='cfp')

plot(ft,
     type='cfpp') 

# Density
plot(ft,
     type='d') 

# Summary
ft

summary(ft)  # the same

print(ft)    # the same

show(ft)     # the same

summary(ft,
        format=TRUE)

summary(ft,
        format=TRUE, 
        pattern='%05.2f')  # regular expression

summary(ft,
        col=c(1:2, 4, 6), 
        format=TRUE,
        pattern='%05.2f')

print(ft,
      col=c(1:2, 4, 6))

print(ft,
      col=c(1:2, 4, 6), 
      format=TRUE,
      pattern='%05.2f')

# Using by
levels(mdf$X1)

plot(fdt(mdf,
         k=5,
         by='X1',
         na.rm=TRUE),
     col=rainbow(5))

levels(mdf$X2)

summary(fdt(iris,
            k=5),
        format=TRUE,
        patter='%04.2f')

plot(fdt(iris,
         k=5),
     col=rainbow(5))

levels(iris$Species)

summary(fdt(iris,
            k=5,
            by='Species'),
        format=TRUE,
        patter='%04.2f')

plot(fdt(iris,
         k=5,
         by='Species'),
     v=TRUE)

#=========================
# Matrices: multivariated
#=========================
summary(fdt(state.x77),
        col=c(1:2, 4, 6),
        format=TRUE)

plot(fdt(state.x77))

# Very big
summary(fdt(volcano,
            right=TRUE),
        col=c(1:2, 4, 6),
        round=3,
        format=TRUE,
        pattern='%05.1f')

plot(fdt(volcano,
         right=TRUE))

## Categorical
x <- sample(x=letters[1:5],
            size=5e2,
            rep=TRUE)

(fdt.c <- fdt_cat(x))

(fdt.c <- fdt_cat(x,
                  sort=FALSE))

#================================================
# Data.frame: multivariated with two categorical
#================================================
mdf <- data.frame(c1=sample(LETTERS[1:3], 1e2, rep=TRUE),
                  c2=as.factor(sample(1:10, 1e2, rep=TRUE)),
                  n1=c(NA, NA, rnorm(96, 10, 1), NA, NA),
                  n2=rnorm(100, 60, 4),
                  n3=rnorm(100, 50, 4),
                  stringsAsFactors=TRUE)

head(mdf)

(fdt.c <- fdt_cat(mdf))

(fdt.c <- fdt_cat(mdf,
                  dec=FALSE))

(fdt.c <- fdt_cat(mdf,
                  sort=FALSE))

(fdt.c <- fdt_cat(mdf,
                  by='c1'))

#================================================
# Matrix: two categorical
#================================================
x <- matrix(sample(x=letters[1:10],
                   size=100,
                   rep=TRUE),
            nc=2,
            dimnames=list(NULL,
                          c('c1', 'c2')))

head(x)

(fdt.c <- fdt_cat(x))