Autocorrelation in project Tycho’s measles data

June 1, 2014

(This article was first published on Wiekvoet, and kindly contributed to R-bloggers)

Project Tycho includes data from all weekly notifiable disease reports for the United States dating back to 1888. These data are freely available to anybody interested.I have looked at Ptoject Tycho’s measles data before, general look, incidence, some high incidence data and correlation between states. After a detour, it is now time to look at the autocorrelations in these data. These show positive correlation at three years.


first steps, see correlation between states

Preparation for autocorrelation

As detailed before, the data contain weekly counts. Summer has less incidence, winter more. For this reason calender year is abolished and a shifted year used (named cycle), which runs from summer to summer. The data, real world as they are, contain plenty of missings. Arbitrarily chosen, if a cycle has data from at least 40 weeks, then I will us this particular year.
r7 <- aggregate(
            sum(x) else

To calculate an autocorrelation, a set of consecutive years are needed. Again arbitrarily chosen, 15 years is the minimum. As a first attempt I just kicked out all missing years. Since that resulted in sufficient states with data, no attempts to refine were made. As additional item the number of data points is stored. All in a nice list.
la <- lapply(levels(r7$State),function(x) {
        datain <- r7[r7$State==x,]
        datain <- datain[complete.cases(datain),]
        if (nrow(datain)==(1+max(datain$cycle)-min(datain$cycle)) &
            aa <- acf(datain$x,plot=FALSE,lag.max=6) else aa <- TRUE

To make a plot, the autocorrelations are pulled out and it is all stuck in a dataframe.
la2 <- la[which(sapply(la,function(x) class(x$aa))==’acf’)]
scfs <-,function(x) as.numeric(x$aa$acf))))
scfs$state <- levels(r7$State)[sapply(la,function(x) class(x$aa)==’acf’)]
scfs$n <- sapply(la2,function(x) x$nr)

And a reshape prior to plotting.
tc <- reshape(scfs,


The plot shows a somewhat negative correlation after one year and a positive correlation after 3 years. The only state not to show the positive correlation after 3 years had much less data, hence for conclusion I ignore that result.

        aes(y=acf, x=lag,group=state,col=state) )+
    geom_line(aes(size = n)) +
    scale_size(range = c(0.5, 3))+


Willem G. van Panhuis, John Grefenstette, Su Yon Jung, Nian Shong Chok, Anne Cross, Heather Eng, Bruce Y Lee, Vladimir Zadorozhny, Shawn Brown, Derek Cummings, Donald S. Burke. Contagious Diseases in the United States from 1888 to the present. NEJM 2013; 369(22): 2152-2158.

To leave a comment for the author, please follow the link and comment on their blog: Wiekvoet. offers daily e-mail updates about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...

If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.


Mango solutions

RStudio homepage

Zero Inflated Models and Generalized Linear Mixed Models with R

Dommino data lab

Quantide: statistical consulting and training



CRC R books series

Six Sigma Online Training

Contact us if you wish to help support R-bloggers, and place your banner here.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)