Your homework must be submitted in R Markdown format. We will not (indeed, cannot) grade homeworks in other formats. Your responses must be supported by both textual explanations and the code you generate to produce your result. (Just examining your various objects in the “Environment” section of R Studio is insufficient—you must use scripted commands.)
We will practice data transformations of a data set consisting of aggregate influenza level measurements in Delaware, District of Columbia, Maryland, Pennsylvania, Virginia, and West Virginia. Using the following code, read in the data frame flumat.
load(url("http://www.stat.cmu.edu/~ryantibs/statcomp-F15/homework/flu.Rdata"))Call head(flumat) to examine the first few rows of flumat. This data frame contains three columns:
wili: weekly influenza occurrences (“wili” stands for weighted influenza like illness percentage) for the these six states, during the time period between 1997 and the first part of 2013. There are some missing values in the first few seasons, don’t panic! This is normal.
season: flu is a seasonal epidemic, and the seasons are defined as week 21 of the current year through the following year’s week 20. Season 1 consists of the commensurate weeks of 1997-1998, season 2 is that of 1998-1999, and so forth.
week: calendar week numbers of each observations.
Notice the seasons are out of order. Reorder the data in increasing order of season. Check that it worked, by producing a plot of the flu measurements over time, and comparing to http://www.stat.cmu.edu/~ryantibs/statcomp-F15/homework/wiliplot.pdf, found on the course website. Label and format your plot legibly.
Say we are interested in flu occurrence in December (week 49 through 52). Extract these rows and call this data frame flumat.12. Display the first ten rows. Using tapply() or some *ply function of your choice, extract the mean December wili in each season, and plot them over time (line plot). Report the seasons with the highest three and lowest three average December flu measurements, and plot them as points in red and blue respectively in the same plot. What is the overall distribution of December averages? The mean and standard deviation of these averages?
flumat is currently arranged in increasing order of season. Rearrange flumat to be arranged by week number, while preserving the increasing order of seasons, and call this new data frame flumat.by.week. Display the first and last 10 rows.
flumat is currently in long format, which means that the data are stacked vertically by season. Sometimes it is convenient to transform such data into wide format with horizontal stacking. In our case, this will mean that each season’s flu occurrences are stacked horizontally as columns – i.e. each column contains a season’s wili values. Do this transformation and call it flumat.wide, and display the first ten rows using head(). Note: the resulting matrix should have 52 rows – the 1st row with 21st calendar week’s measurements, the second row with that of week 22, and so forth. (Hint: probably the most straightforward way of doing this is to use a for loop, and fill in the wide matrix one row at a time, by week. A more streamlined, but also more advanced strategy, would be to use the reshape() function in R.)
Bonus: Already loaded in your computer is new partially observed flu occurrence for season 17 – flumat.new – that is formatted in the same way as flumat. Merge it with the wide-formatted flumat.wide using merge() while preserving the formatting details of flumat.wide. Display the first six rows of the new data frame. (Hint: You may want to make a new column in flumat.wide and use it to align the merge.)
In this part of the homework you will debug several pieces of almost-functional code. Start by setting eval=TRUE for a block of code, then use debugging techniques developed in class to find any problems in that block. After correcting those problems, explain in words what changes you made and why. Your final submission should have the setting eval=TRUE, and everything should run.
zero.finder(). Be sure that zero.finder provides useful error messages when it encounters a problem. Hint: add extra intermediate outputs to diagnose what the function is doing.zero.finder <- function (fn, start, tol=1e-6, delta=1e-4) {
pp <- start
repeat {
grad.fn <- (fn(pp+delta) - fn(pp-delta))/(2*delta)
pp <- pp + fn(p)/grad.fn
if (abs(fn(pp)) < tol) break
}
pp
}
zero.finder (function(x) {1 - x^2}, 1.5) # should be 1
zero.finder (function(x) {1 - x^2}, 0.5) # should be 1
zero.finder (function(x) {x^3 - 7*x^2 + 7*x + 15}, 3.5) # should be 3
zero.finder (function(x) {x^3 - 7*x^2 + 7*x + 15}, -1.5) # should be -1
zero.finder (function(x) {x^3 - 7*x^2 + 7*x + 15}, 4.5) # should be 5
zero.finder (function(x) {1/(1-x^2)}, 1) # what should happen here?dynamic.fibonacci(). It should return the numbers corresponding to 1,1,2,3,5,8,13 ... for those values, and NA for any disallowed values.dynamic.fibonacci <- function (nn) {
if ( !(is.integer(nn)) | (nn<1)){
print ("Bad input")
return (NA)
}
my.fib <- c(1,1)
for (kk in 3:nn) my.fib[kk] <- my.fib[kk-1] + my.fib[kk-2]
return my.fib[nn]
}
dynamic.fibonacci(7) #Should be 13
dynamic.fibonacci(5) #Should be 5
dynamic.fibonacci(1) #Should be 1
dynamic.dibonacci(1.5) #Should be NA
dynamic.dibonacci(0) #Should be NAflumat.copy = flumat #We'll work on a copy of the data
week50 = (flumat.copy$week = 50) #The first boolean vector
week51 = (flumat.copy$week = 51) #The second boolean vector
week50wili = flumat.copy$wili[week50,]
week51wili = flumat.copy$wili[week51,]
flureg = lm(week51wili~week50wili)