rata: An Imperative Library for Data Manipulation

rm(list=ls(all=TRUE))
library(genvar)
capture(clear(), silent=TRUE)
library(plm)
data(Produc)

## "use" loads the data set Produc into memory
use(Produc)
listif()

## preserve data set
p = preserve()

## list variables in dataset
describe()

## sum over emp by year
collapse("sum(emp) sum(unemp)", year)
listif()

## restore original data
restore(p, replace=TRUE)

## reshape dataset from (state,year,emp) to (state,emp1970,emp1971,...)
shape(state~emp|year, direction="wide")

## listif(expr) prints the dataset if the statement is true (it also returns the part of the dataset
## that satisfies the condition)
listif()

## reshape dataset from (state, emp1970, emp1971,...) to (state,year,emp)
shape(state~year|emp, direction="long")
listif()

## list emp by year just for Wyoming
listif(state == "WYOMING")

describe()
## year is a character so destring it
destring(year)


## generate emp*year for no reason
gen(empyear, emp*year)
listif()

count()
count(emp <= 1000)

## test out regression
use(Produc, clear=TRUE)

r = reg(emp, unemp)
r

xtset(year, state)
xtbalanced()

r = reg(emp, unemp, hac=year)
r
r = reg(emp, unemp, cluster=year)
r
p = preserve()
keepvar("state year emp unemp")

## add some garbage data to show addobs, not real data
addobs("state='Mars',year=1990,emp=100,unemp=4.0")
fillin("state year")

r = reg(emp, unemp, effect="twoways", cluster=year)
r

restore(p, replace=TRUE)

## Binary regression

gen(empmedian, emp > median(emp))

r = logit(empmedian, unemp)
r

## Show a graph of fraction employment over time

### Prepare data

gen(laborforce, emp/(1-unemp/100))
empfrac = function(emp, laborforce) sum(emp)/sum(laborforce)
collapse("empfrac(emp,laborforce)", year)
rename(empfrac(emp, laborforce), empfrac)
destring(year)

### Plot data
gvplot(year, empfrac, type="b", main="Employment Percentage over Time",
       xlab="Year", ylab="Employment Percentage", pch=19)

## Can use getdata to pull the data frame from genvar to the R environment
data = getdata()