misc_stata_ados

Misc Utility programs in Stata. Brief intros below.

discretize

Creates discrete values (bins) for a specified continuous variable, either using the percentile cutpoints specified in cutpoints(a, b, c) or into N number of uniform sized bins as specified in nbins(n). Useful when trying to frame a regression specification as a classification problem to be handled using an ordered/multinomial logit (e.g. low / medium / high cost based on cutpoints).

discretize total_cost, gen(cost_level) cut(25 50 75)
discretize total_cost, gen(bins) nbins(200)

winsorize

Winsorizes specified variable at cutpoints specified in AT(lowerbound upperbound) or lim(limit 100-limit) and optionally generates new variable.

winsorize price, gen(newprice) at (1 99)

freq_table

Replaces dataset in memory with a frequency table of variables and interactions. Accepts dummy variables, factor variables, and their interactions and produces a labelled table (by extracting appropriate variable and value labels, if they exist) of counts for dummies (e.g. female, rur_urb ), each level of factor variables (i.education, i.country) and each cell in the crosstab between categorical variables separated by * or # (i.education#i.country).

Example of use:

  use exampledata, clear // contains individual level data on income, sex, education, country, rural/urban location
  gl rhs_vars female rur_urb i.educ i.country i.education#i.country
  preserve
  freq_table $rhs_vars
  save freqs, replace
  restore

freqs.dta now contains:

Raw	Label	Count	Pct
rur_urb == 1	Urban == 1	24	0.2
educ == 1	Education == No HS	43	0.36
educ == 2	Education == HS	40	0.33
educ == 3	Education == College	24	0.2
educ == 1 X country == 2	Education == No HS X Country == United States	12	0.1

and so on.

dot_product

Calculates the variable Y = XB where X is a subset of N variables in the currently loaded dataset, B is an arbitrary column vector (NX1 matrix). Basically a way to construct predicted values from a regression when the coefficients have been stored in a matrix / read in from elsewhere. Produces identical results to predict when used with the postestimation e(b) coefficient vector.

sysuse auto, clear
mat A = [1\2\3]
dot_product fitted_val A price weight trunk

prefix_labels

Adds prefix of variable label / variable name to stata value labels so that regression output can be filtered and sorted in excel. So, value labels for values 1 "United States" 2 "Nepal" 3 "United Kingdom" become 1 "Country: United States" 2 "Country: Nepal" 3 "Country: United Kingdom" , so that excel's filter and sort functions work nicely.

  use exampledata, clear // contains individual level data on income, sex, education, country, rural/urban location
  prefix_labels sex country education
  reg income sex education
  esttab using "output.csv", label replace

bettertab

Wrapper for default tab/tab2 commands that temporarily adds numeric value prefixes and drops them afterwards (so that they don't affect graphs etc.)

bettertab race sex

returns

Race	1.F	2.M	Total
1. Black	1	2	3
2. White	4	5	9
3. Asian	7	8	15
4. Native American	10	11	21

count_unique

Duplicate functionality with codebook, but returns scalar that can be used for calculations / stored as a variable in a loop.

count_unique teacher classroom
sca ntc = `r(nv)'

duprep

Detailed report on duplicates / missing values in variable.

duprep student_id
// returns
/*
*______student_id___________*
Distinct populated obs : 542
% Singletons : 45
Min obs : 1
Mean obs : 4
Max obs: 50
% of obs with missing values: 1
*/

dtimer

A display-friendly wrapper of the default timer that displays runtime of any section of code between dtimer on and dtimer off in hours/minutes/seconds.

lookin

Searches for string specified in for() in varlist, optionally generates flag for observations where matches were found.

lookin enr2000 enr2001 enr2002, for("Y") g(enr_2000_2002)

unstable

Checks for variation in variable(s) across other variable(s)

unstable gender age, by(student)

partition_var

Takes variable and cutpoints and generates dummies with prefix specified in prefix. Example:

partition_var age, cut(0 35 50 75) prefix(age)

generates the variables (with the appropriate variable labels): a_0_35 a_36_50 a_51_75 a76

pathmake

Generates entire folder structure for path necessary, which the native mkdir command cannot do.

pathmake "C:/Users/alal/Desktop/test1/temp/test2/test3/test4/test5"

creates the entire folder structure, even though the subdirectories didn't exist to begin with.

cond_stitcher

Returns a long string separated by OR (|) or AND(&) operators that can be used in subsequent calculations.

loc test "age05 age610 age1115 male old"
cond_stitcher `test', sep(|)
// returns "age05|age610|age1115|male|old"
count if `r(cond)'
> 55

ds2

Wrapper for ds command that does not abbreviate variable names. Preferable to ds for interactive use.

okeep

Order and Keep varlist.

Installation

Run the following line in the Stata console:

net install lal_utilities, from(https://raw.github.com/apoorvalal/misc_stata_ados/master/)

Or, if you prefer, download ados and move to your personal ado folder / c(sysdir_personal) (where ssc-installed ados live) Will upload sthlp files at some point.