Anda di halaman 1dari 8

Histogram

From Wikipedia, the free encyclopedia


For the histograms used in digital image processing, see Image histogram and Color
histogram.
Histogram
First
described
by
Karl Pearson
Purpose
To roughly assess the probability
distribution of a given variable by
depicting the freuencies of observations
occurring in certain ranges of values
In statistics, a histogram is a graphical representation of the distribution of data. It is an
estimate of the probability distribution of a continuous variable and !as first introduced
by Karl Pearson.
"#$
% histogram is a representation of tabulated freuencies, sho!n as
ad&acent rectangles or suares 'in some situations(, erected over discrete intervals 'bins(,
!ith an area proportional to the freuency of the observations in the interval. The height
of a rectangle is also eual to the freuency density of the interval, i.e., the freuency
divided by the !idth of the interval. The total area of the histogram is eual to the
number of data. % histogram may also be normali)ed displaying relative freuencies. It
then sho!s the proportion of cases that fall into each of several categories, !ith the total
area eualing #. The categories are usually specified as consecutive, non*overlapping
intervals of a variable. The categories 'intervals( must be ad&acent, and often are chosen
to be of the same si)e.
"+$
The rectangles of a histogram are dra!n so that they touch each
other to indicate that the original variable is continuous.
",$
-istograms are used to plot the density of data, and often for density estimation.
estimating the probability density function of the underlying variable. The total area of a
histogram used for probability density is al!ays normali)ed to #. If the length of the
intervals on the x*a/is are all #, then a histogram is identical to a relative freuency plot.
%n alternative to the histogram is kernel density estimation, !hich uses a kernel to
smooth samples. This !ill construct a smooth probability density function, !hich !ill in
general more accurately reflect the underlying variable.
The histogram is one of the seven basic tools of uality control.
"0$
Contents
# 1tymology
+ 1/amples
, 2athematical definition
o ,.# Cumulative histogram
o ,.+ 3umber of bins and !idth
0 4ee also
5 6eferences
7 Further reading
8 1/ternal links
Etymology
%n e/ample histogram of the heights of ,# 9lack Cherry trees.
The etymology of the !ord histogram is uncertain. 4ometimes it is said to be derived
from the :reek histos ;anything set upright; 'as the masts of a ship, the bar of a loom, or
the vertical bars of a histogram(< and gramma ;dra!ing, record, !riting;. It is also said
that Karl Pearson, !ho introduced the term in #=>#, derived the name from ?historical
diagram?.
"5$
Examples
The @.4. Census 9ureau found that there !ere #+0 million people !ho !ork outside of
their homes.
"7$
@sing their data on the time occupied by travel to !ork, Table + belo!
sho!s the absolute number of people !ho responded !ith travel times ?at least ,A but
less than ,5 minutes? is higher than the numbers for the categories above and belo! it.
This is likely due to people rounding their reported &ourney time.
"citation needed$
The problem
of reporting values as some!hat arbitrarily rounded numbers is a common phenomenon
!hen collecting data from people.
"citation needed$
-istogram of travel time 'to !ork(, @4 +AAA census. %rea under the curve euals the total
number of cases. This diagram uses BC!idth from the table.
Data by absolute numbers
Interval Width Quantity Quantity/width
A 5 0#=A =,7
5 5 #,7=8 +8,8
#A 5 #=7#= ,8+,
#5 5 #>7,0 ,>+7
+A 5 #8>=# ,5>7
+5 5 8#>A #0,=
,A 5 #7,7> ,+8,
,5 5 ,+#+ 70+
0A 5 0#++ =+0
05 #5 >+AA 7#,
7A ,A 707# +#5
>A 7A ,0,5 58
This histogram sho!s the number of cases per unit interval as the height of each block, so
that the area of each block is eual to the number of people in the survey !ho fall into its
category. The area under the curve represents the total number of cases '#+0 million(.
This type of histogram sho!s absolute numbers, !ith B in thousands.
-istogram of travel time 'to !ork(, @4 +AAA census. %rea under the curve euals #. This
diagram uses BCtotalC!idth from the table.
Data by proportion
Interval Width Quantity Q! Q/total/width
A 5 0#=A A.AA78
5 5 #,7=8 A.A++#
#A 5 #=7#= A.A,AA
#5 5 #>7,0 A.A,#7
+A 5 #8>=# A.A+>A
+5 5 8#>A A.A##7
,A 5 #7,7> A.A+70
,5 5 ,+#+ A.AA5+
0A 5 0#++ A.AA77
05 #5 >+AA A.AA0>
7A ,A 707# A.AA#8
>A 7A ,0,5 A.AAA5
This histogram differs from the first only in the vertical scale. The area of each block is
the fraction of the total that each category represents, and the total area of all the bars is
eual to # 'the fraction meaning ?all?(. The curve displayed is a simple density estimate.
This version sho!s proportions, and is also kno!n as a unit area histogram.
In other !ords, a histogram represents a freuency distribution by means of rectangles
!hose !idths represent class intervals and !hose areas are proportional to the
corresponding freuencies. the height of each is the average freuency density for the
interval. The intervals are placed together in order to sho! that the data represented by
the histogram, !hile e/clusive, is also contiguous. '1.g., in a histogram it is possible to
have t!o connecting intervals of #A.5E+A.5 and +A.5E,,.5, but not t!o connecting
intervals of #A.5E+A.5 and ++.5E,+.5. 1mpty intervals are represented as empty and not
skipped.(
"8$
"athematical de#inition
%n ordinary and a cumulative histogram of the same data. The data sho!n is a random
sample of #A,AAA points from a normal distribution !ith a mean of A and a standard
deviation of #.
In a more general mathematical sense, a histogram is a function mi that counts the number
of observations that fall into each of the dis&oint categories 'kno!n as bins(, !hereas the
graph of a histogram is merely one !ay to represent a histogram. Thus, if !e let n be the
total number of observations and k be the total number of bins, the histogram mi meets
the follo!ing conditions.
Cumulative histogram
% cumulative histogram is a mapping that counts the cumulative number of observations
in all of the bins up to the specified bin. That is, the cumulative histogram Mi of a
histogram mj is defined as.
$umber o# bins and width
There is no ?best? number of bins, and different bin si)es can reveal different features of
the data. :rouping data is at least as old as :raunt;s !ork in the #8th century, but no
systematic guidelines !ere given
"=$
until 4turges;s !ork in #>+7.
">$
@sing !ider bins !here the density is lo! reduces noise due to sampling randomness<
using narro!er bins !here the density is high 'so the signal dro!ns the noise( gives
greater precision to the density estimation. Thus varying the bin*!idth !ithin a histogram
can be beneficial. 3onetheless, eual*!idth bins are !idely used.
4ome theoreticians have attempted to determine an optimal number of bins, but these
methods generally make strong assumptions about the shape of the distribution.
Depending on the actual data distribution and the goals of the analysis, different bin
!idths may be appropriate, so e/perimentation is usually needed to determine an
appropriate !idth. There are, ho!ever, various useful guidelines and rules of thumb.
"#A$
The number of bins k can be assigned directly or can be calculated from a suggested bin
!idth h as.
The braces indicate the ceiling function.
4uare*root choice
!hich takes the suare root of the number of data points in the sample 'used by 1/cel
histograms and many others(.
"##$
4turges; formula
4turges; formula
">$
is derived from a binomial distribution and implicitly assumes an
appro/imately normal distribution.
It implicitly bases the bin si)es on the range of the data and can perform poorly if n F ,A.
"citation needed$
It may also perform poorly if the data are not normally distributed.
6ice 6ule
The 6ice 6ule
"#+$
is presented as a simple alternative to 4turges;s rule.
Doane;s formula
Doane;s formula
"#,$
is a modification of 4turges; formula !hich attempts to improve its
performance !ith non*normal data.
!here is the estimated ,rd*moment*ske!ness of the distribution and
4cott;s normal reference rule
!here is the sample standard deviation. 4cott;s normal reference rule
"#0$
is optimal for
random samples of normally distributed data, in the sense that it minimi)es the integrated
mean suared error of the density estimate.
"=$
FreedmanEDiaconis; choice
The FreedmanEDiaconis rule is.
"#5$"=$
!hich is based on the interuartile range, denoted by IB6. It replaces ,.5G of 4cott;s rule
!ith + IB6, !hich is less sensitive than the standard deviation to outliers in data.
Choice based on minimi)ation of an estimated L
+"#7$
risk function
!here and are mean and biased variance of a histogram !ith bin*!idth ,
and .
6emark
% good reason !hy the number of bins should be proportional to is the follo!ing.
suppose that the data are obtained as independent reali)ations of a bounded probability
distribution !ith smooth density. Then the histogram remains eually HruggedI as tends
to infinity. If is the H!idthI of the distribution 'e. g., the standard deviation or the inter*
uartile range(, then the number of units in a bin 'the freuency( is of order and the
relative standard error is of order . Comparing to the ne/t bin, the relative
change of the freuency is of order provided that the derivative of the density is non*
)ero. These t!o are of the same order if is of order , so that is of order .
This simple cubic root choice can also be applied to bins !ith non*constant !idth.
%ee also
Wikimedia Commons has media related to Histograms.
Statistics portal
Data binning
Density estimation
o Kernel density estimation, a smoother but more comple/ method of
density estimation
FreedmanEDiaconis rule
Image histogram
Pareto chart
4even 9asic Tools of Buality
J*optimal histograms

Anda mungkin juga menyukai