Secretary of Education Report

January 24, 2014

For more info also see this post.

Grab the Data

URL <- "http://en.wikipedia.org/wiki/United_States_Secretary_of_Education" 
Table2 <- readHTMLTable(URL,  
    colClasses = rep("character", 2), 
    skip.rows=1, 
    which=2)

tab <- Table2[-4, -c(1,2,4,7 )] 

tab[, 4] <- as.character(tab[, 4])
tab[2, 4] <- tab[1, 4]
tab[4, 4] <- "George H. Bush"
tab[3, 4] <- paste(tab[c(2, 4), 4], collapse="/")
tab[7, 4] <- tab[6, 4]
tab[] <- lapply(tab, as.character)

tab <- rbind(c("Shirley M. Hufstedler", "November 30, 1979", 
    "January 20, 1981", "Jimmy Carter"), tab)
tab$k_12_educator <- as.logical(c(0, 1, 0, 0, 0, 0, 0, 0, 0))
names(tab)[1:4] <- c("Secretary_of_Ed.", "Took_Office", "Left_Office", "President")
rownames(tab)<-NULL
tab[,3] <- as.Date(tab[,3], format = "%B %d, %Y")
tab[,2] <- as.Date(tab[,2], format = "%B %d, %Y")
tab[9, 3] <- Sys.Date()
tab$days <- as.numeric(difftime(strptime(tab[,3], format = "%Y-%m-%d"), 
    strptime(tab[, 2], format = "%Y-%m-%d")))
tab$President_Party <- c("Republican")
tab$President_Party[c(1, 6, 9)] <- "Democrat"

The Secretaries of Education

tab2 <- tab
tab2$days <- round(tab2$days, 0)
tab2[] <- lapply(tab2, as.character)
print(xtable(tab2), type="html")

	Secretary_of_Ed.	Took_Office	Left_Office	President	k_12_educator	days	President_Party
1	Shirley M. Hufstedler	1979-11-30	1981-01-20	Jimmy Carter	FALSE	417	Democrat
2	Terrel H. Bell	1981-01-22	1985-01-20	Ronald Reagan	TRUE	1459	Republican
3	William J. Bennett	1985-02-06	1988-09-20	Ronald Reagan	FALSE	1322	Republican
4	Lauro F. Cavazos	1988-09-20	1990-12-12	Ronald Reagan/George H. Bush	FALSE	813	Republican
5	Lamar Alexander	1991-03-22	1993-01-20	George H. Bush	FALSE	670	Republican
6	Richard W. Riley	1993-01-21	2001-01-20	Bill Clinton	FALSE	2921	Democrat
7	Roderick R. Paige	2001-01-20	2005-01-20	George W. Bush	FALSE	1461	Republican
8	Margaret Spellings	2005-01-20	2009-01-20	George W. Bush	FALSE	1461	Republican
9	Arne Duncan	2009-01-21	2014-01-24	Barack Obama	FALSE	1829	Democrat

tab2 <- tbl_df(tab)
time <- sum(tab2$days)
stats1 <- tab2 %.%
    group_by(k_12_educator) %.%
    summarise(time_by_educ_status = sum(days), Per_time_by_party = 100*time_by_educ_status/time)

stats2 <- tab2 %.%
    group_by(President_Party, k_12_educator) %.%
    summarise(time_by_party = sum(days), Per_time_by_party = 100*time_by_party/time)

ggplot(stats1, aes(x=factor(1), fill=k_12_educator)) + 
    geom_bar(aes(weight=Per_time_by_party), position="dodge") +
    scale_fill_manual(values=c("grey25", "red"), name= "K - 12\nEducator") +
    ylab("Percent of Time") + xlab("K-12 Educator") + theme_basic() +
    theme(axis.text.x=element_blank(), axis.ticks.x=element_blank()) +
    scale_y_continuous(expand = c(0, 0), limits = c(0, 90))

plot of chunk unnamed-chunk-3

print(xtable(stats1), type="html")

	k_12_educator	time_by_educ_status	Per_time_by_party
1	FALSE	10894.00	88.19
2	TRUE	1459.00	11.81

x <- data.frame("Democrat", TRUE, 0, 0)
names(x) <- names(stats2)
stats3 <- rbind(stats2, x)
ggplot(stats3, aes(x=k_12_educator, fill=k_12_educator)) + 
    geom_bar(aes(weight=Per_time_by_party), position="dodge", size=2) +
    ylab("Percent of Time") + xlab("K-12 Educator") +
    facet_grid(~President_Party) +
    scale_fill_manual(values=c("grey25", "red"), name= "K - 12\nEducator") +
    theme_basic() +
    scale_y_continuous(expand = c(0, 0), limits = c(0, 50))

plot of chunk unnamed-chunk-4

print(xtable(stats2), type="html")

	President_Party	k_12_educator	time_by_party	Per_time_by_party
1	Democrat	FALSE	5167.00	41.83
2	Republican	FALSE	5727.00	46.36
3	Republican	TRUE	1459.00	11.81