R Notebook

library(swirl)
swirl()

Explore the data

dim(pm0)
head(pm0)

Making the first row of names

cnames<- strsplit(cnames, "|", fixed = T)
names(pm0)<-make.names(cnames[[1]][wcol])

The measurements of particulate matter (pm25) are in the column named Sample.Value. Assign this component of pm0 to the variable x0

x0<-pm0$Sample.Value

What percentage of values are missing in this vector?

mean(is.na(x0))

Same actions for 2012 data

names(pm1)<-make.names(cnames[[1]][wcol])
x1<-pm1$Sample.Value
mean(is.na(x1))

summary(x0)
summary(x1)

boxplot(x0,x1)
boxplot(log10(x0), log10(x1))

AMount of negative values

negative <- x1<0
sum(negative, na.rm = TRUE)
mean(negative, na.rm = TRUE)

When does the negative data occure?

dates<-pm1$Date
dates<-as.Date(as.character(dates), "%Y%m%d")
hist(dates[negative], "month")

Biding county and monitors together, We see that 10 monitors in New York State were active in both 1999 and 2012.

str(site0)
str(site1)
both<-intersect(site0, site1)

Creating var with NY county and monitors

cnt0 <- subset(pm0, State.Code == 36 & county.site %in% both)
cnt1 <- subset(pm1, State.Code == 36 & county.site %in% both)

This will split cnt0 into several data frames according to county.site (that is, monitor IDs) and tell us how many measurements each monitor recorded.

sapply(split(cnt0, cnt0$county.site), nrow)
sapply(split(cnt1, cnt1$county.site), nrow)

We want to examine a monitor with a reasonable number of measurements so let’s look at the monitor with ID 63.2008.

pm0sub<-subset(cnt0, County.Code== 63 & Site.ID == 2008 )
pm1sub<-subset(cnt1, County.Code== 63 & Site.ID == 2008 )

Now we’d like to compare the pm25 measurements of this particular monitor (63.2008) for the 2 years.

x0sub<- pm0sub$Sample.Value
x1sub<- pm1sub$Sample.Value

We’d like to make our comparison visually so we’ll have to create a time series of these pm25 measurements

dates0<-as.Date(as.character(pm0sub$Date),"%Y%m%d" )
dates1<-as.Date(as.character(pm1sub$Date),"%Y%m%d" )

plotting the difference

par(mfrow=c(1,2), mar = c(4,4,2,1))
plot(dates0, x0sub, pch = 20)
abline(h = median(x0sub, na.rm = TRUE),lwd=2)
plot(dates1, x1sub, pch = 20)
abline(h = median(x1sub, na.rm = TRUE),lwd=2)
rng<- range(x0sub, x1sub, na.rm=TRUE)

a comparison of all the states’ mean pollution levels.

mn0 <- with(pm0,tapply(Sample.Value,State.Code,mean,na.rm=TRUE))
mn1 <- with(pm1,tapply(Sample.Value,State.Code,mean,na.rm=TRUE))

dim(mrg)

[1] 52  3

| Keep working like that and you'll get there!

  |                                                                    
  |=======================================================       |  89%

| We see merge has 52 rows and 3 columns. Since the Virgin Island data
| was missing from d1, it is excluded from mrg. Look at the first few
| entries of mrg using the head command.

Plotting the results

with(mrg, plot(rep(1, 52), mrg[, 2], xlim = c(.5, 2.5)))#points

| Not quite, but you're learning! Try again. Or, type info() for more
| options.

| Type mrg[mrg$mean.x < mrg$mean.y,] at the command prompt.

with(mrg, points(rep(2, 52), mrg[, 3]))#points

| One more time. You can do it! Or, type info() for more options.

| Type mrg[mrg$mean.x < mrg$mean.y,] at the command prompt.

segments(rep(1, 52), mrg[, 2], rep(2, 52), mrg[, 3])#connect the dots

| That's not exactly what I'm looking for. Try again. Or, type info() for
| more options.

| Type mrg[mrg$mean.x < mrg$mean.y,] at the command prompt.

mrg[mrg$mean.x < mrg$mean.y,]