3.59 msummary Calculate Statistics for 1 Variable

Calculate the type of statistics specified at c= parameter for fields specified at f= parameter.

Format

msummary c= f= [a=] [k=] [i=] [o=] [-nfn] [-nfno] [-x] [-q] [precision=] [--help] [--version]

Parameters

`k=`	Compute statistics based on the key field(s) specified (multiple fields can be specified).
`f=`	Field lists for computation of statistical summary (multiple fields can be specified).
	When `-x`,`-nfn` option is specified, specify the field number (0 ).
`c=`	Statistics (multiple fields can be specified)
	Specify list of statistics delimited by comma.
	Statistics list:
	`sum/mean/count/ucount/devsq/var/uvar/sd/usd/cv/min/qtile1/median/qtile3/max/`
	`range/qrange/mode/skew/uskew/kurt/ukurt`
`-a`	New column name.
	Results from calculation on column(s) specified at `f=` parameter (default is fld).

List of Statistics

The list of statistics specified at c= parameter is shown in Table 3.33.

Table 3.33: List of Statistics

Value of `c=`	Description	Equation	Remarks
count	Count (Except NULL value)	$n$ : Number of non-NULL records	It can not be applied to character string field.
ucount	Unique count	$un$ : Number of duplicate values removed	It can not be applied to character string field.
sum	Total	$sum=\sum _{i=1}^ n x_ i$
mean	Arithmetic mean	$m=\frac{1}{n}\sum _{i=1}^ n x_ i$
devsq	Sum of squared deviation	$S=\sum _{i=1}^ n(x_ i-m)^2$
var	Variance	$s^2=\frac{1}{n}S$
uvar	Variance (unbiased estimate)	$u^2=\frac{1}{n-1}S$
sd	Standard deviation	$s=\sqrt {s^2}$
usd	Standard deviation (sort of unbiased variance)	$u=\sqrt {u^2}$	commonly used standard deviation
cv	Coefficient of variation	$cv=s/m100\%$
mode	Mode	$mode$ : Most frequent value	Print the value of the smaller value
			if the frequency is same.
min	Minimum value	$min=\min _ i x_ i$
max	Maximum value	$max=\max _ i x_ i$
range	Range	$r=max-min$
median	Median	$Q2=Second quartile when sorted in ascending order$
qtile1	First quartile	$Q1=First quartile when sorted in ascending order$
qtile3	Third quartile	$Q3=Third quartile when sorted in ascending order$
qrange	Interquartile range	$rq=Q3-Q1$
skew	Skewness	$\frac{\frac{1}{n}\sum _{i=1}^ n (x_ i-m)^3}{s^3}$
uskew	Skewness (unbiased estimate)	omitted
kurt	Kurtosis	$\frac{\frac{1}{n}\sum _{i=1}^ n (x_ i-m)^4}{s^4}-3.0$
ukurt	Kurtosis (unbiased estimate)	omitted

Examples

Example 1: Basic Example

Find out the median and average "quantity" and "amount" by each customer. Save the output in a new column named "type".

$ more dat1.csv
customer,quantity,amount
A,1,10
A,2,20
B,1,15
B,3,10
B,1,20
$ msummary k=customer f=quantity,amount c=median:medianval,mean:meanval a=type i=dat1.csv o=rsl1.csv
#END# kgsummary a=type c=median:medianval,mean:meanval f=quantity,amount i=dat1.csv k=customer o=rsl1.csv
$ more rsl1.csv
customer%0,type,medianval,meanval
A,quantity,1.5,1.5
A,amount,15,15
B,quantity,1,1.666666667
B,amount,15,15

Related Commands

mstats : Compute one type of statistics.