Lalonde-Matching-Shajara

install.packages("Matching")
library(Matching)


#QUESTION 1:
#seperates our data based on who recieved treatment
control <- nsw_dw[nsw_dw$treat == 0,]
treatment <- nsw_dw[nsw_dw$treat == 1,]


#calculates and prints difference of means
diff_of_means <- mean(treatment$re78) - mean(control$re78)
print(diff_of_means)


linear_reg <- lm(re78 ~ treat, data=nsw_dw)
#will show coefficients
print(linear_reg)
summary(linear_reg)
#confint function can find confidence interval
confint(linear_reg)

#Question 2:

#calculates and prints difference of means
diff_of_means <- mean(treatment$re78) - mean(cps_controls$re78)
print(diff_of_means)

combined_data <- rbind(treatment,cps_controls)
linear_reg2 <- lm(re78 ~ treat, data=combined_data)
#will show coefficients
print(linear_reg2)
summary(linear_reg2)
#confint function can find confidence interval
confint(linear_reg2)

#QUESTION 3
match_comb<-glm(treat ~ age + education + black + hispanic + married + nodegree+
                  re74 + re75, family=binomial(logit), data=combined_data)

D<-combined_data$treat #treatment
Y<-combined_data$re78 #outcome
X  <- match_comb$fitted

#propensity score matching estimator
r1  <- Match(Y=Y, Tr=D, X=X, M=1)
summary(r1)

#check for balance before and after matching
mb1  <- MatchBalance(treat ~ age + education + black + hispanic + married + nodegree+
                       re74 + re75, match.out=r1,
                     data=combined_data, nboots = 20)

#QUESTION 4

full_final_data <- cbind(combined_data, match_comb$fitted.values)

names(full_final_data) <- c("data_id", "treat", "age", "education","black",
                            "hispanic", "married", "nodegree", "re74", "re75",
                            "re78", "fitted_values")

D2<-full_final_data$treat #treatment
Y2<-full_final_data$re78 #outcome
X2 <- cbind(full_final_data$age, full_final_data$education, full_final_data$black,
                          full_final_data$hispanic, full_final_data$married, full_final_data$nodegree,
                          full_final_data$re74, full_final_data$re75, full_final_data$fitted_values)

#propensity score matching estimator
r2  <- Match(Y=Y2, Tr=D2, X=X2, M=1, sample = TRUE)
summary(r2)

#check for balance before and after matching
mb1  <- MatchBalance(treat ~ age + education + black + hispanic + married + nodegree+
                       re74 + re75, match.out=r1,
                     data=full_final_data, nboots = 20)

#QUESTION 5 Part 1 (Almost there!)
install.packages("rbounds")
install.packages("rgenoud")
library(rbounds)
library("rgenoud")

X3 = cbind(full_final_data$age, full_final_data$education, full_final_data$black,
          full_final_data$hispanic, full_final_data$married, full_final_data$nodegree,
          full_final_data$re74, full_final_data$re75)

BalanceMat <- cbind(full_final_data$age, full_final_data$education, full_final_data$black,
                    full_final_data$hispanic, full_final_data$married, full_final_data$nodegree,
                    full_final_data$re74, full_final_data$re75)

genout <- GenMatch(Tr= full_final_data$treat, X=X3, BalanceMatrix=BalanceMat, estimand="ATT", M=1,
                   pop.size=16, max.generations=10, wait.generations=1)

Y3=full_final_data$re78

mout <- Match(Y=Y3, Tr=full_final_data$treat, X=X3, estimand="ATT", Weight.matrix=genout)
summary(mout)

mb <- MatchBalance(full_final_data$treat~full_final_data$age +full_final_data$educ+full_final_data$black+
                     full_final_data$hisp+ full_final_data$married+ full_final_data$nodegree+ full_final_data$re74+ full_final_data$re75,
                   match.out=mout, nboots=20)

mean()
#QUESTION 5 Part 2 (Woohoo!)

X4 = cbind(full_final_data$age, full_final_data$education, full_final_data$black,
          full_final_data$hispanic, full_final_data$married, full_final_data$nodegree,
          full_final_data$re75, full_final_data$re74, full_final_data$propensities)

BalanceMat <- cbind(full_final_data$age, full_final_data$education,
                    full_final_data$black, full_final_data$hispanic,
                    full_final_data$married, full_final_data$nodegree,
                    full_final_data$re75, full_final_data$re74,
                    full_final_data$propensities)

genout <- GenMatch(Tr=full_final_data$treat, X=X, BalanceMatrix=BalanceMat, estimand="ATT", M=1)

Y4 = full_final_data$re78

mout <- Match(Y=Y, Tr=treat, X=X, estimand="ATT", Weight.matrix=genout)
summary(mout)

mb <- MatchBalance(full_final_data$treat~full_final_data$age + full_final_data$education +
                     full_final_data$black + full_final_data$hispanic + full_final_data$married
                   + full_final_data$nodegree,
                   match.out=mout, nboots=20)