Lecture 9 Writing functions.Rmd

---
title: "Lecture 9 Writing functions"
author: "David A McAllister"
date: "Monday 31st October, 2016"
output:
  slidy_presentation: default
---


## Objective of section
* To see how a function is just like interactive R work

## "Don't repeat yourself"
### Andy Hunt and Dave Thomas

## Use functions to avoid repeating yourself

* Reduces chances of error
* Makes code easier to change
* Makes code more readable (once you get your eye in)
* Makes code more transferrable

## Problem
Matrix 10 x 5. 10 hospitals and 5 timepoints.

Do any of the hospitals/timepoints have a high mortality?

Expected mortality is 100.

Want to do a statistical test on each cell.

## Our data
```{r}
set.seed(1234)
hospitals <- matrix(rpois(50,120), nrow = 10, 
                    dimnames = list(paste("hosp", 1:10), paste("t", 1:5)))
hospitals[cbind (c(2,3,4),c(1,2,3))] <- NA
hospitals
```

## Simplify the problem
```{r}
hospital <- hospitals[1,1, drop = FALSE]
```

## Work interactively
```{r}
# poisson.test (x = hospital, r = 100)
poisson.test (x = as.vector(hospital), r = 100)
```

## ?poisson.test
Value
A list with class "htest" containing the following components:

component | explanation
---------- | ------------
statistic | the number of events (in the first sample if there are two.)
parameter | the corresponding expected count
p.value	| the p-value of the test.
conf.int | a confidence interval for the rate or rate ratio.
estimate | the estimated rate or rate ratio.
null.value | the rate or rate ratio under the null, r.
alternative | a character string describing the alternative hypothesis.
method | the character string "Exact Poisson test" or "Comparison of Poisson rates" as appropriate.
data.name | a character string giving the names of the data.

## So using subsetting
```{r}
poisson.test (x = as.vector(hospital), r = 100)$p.value
```

## Wrap it up in a function
``` {r, warning=TRUE}
MyPois <- function (x, ...){
  x <- as.vector(x)
  poisson.test (x, r = 100)$p.value
}
# test it
MyPois(hospitals[1,1])
MyPois(hospitals[1,2])
# MyPois(hospitals[2,1])
```

## Re-write the function
``` {r}
MyPois <- function (x, ...){
  x <- as.vector(x)
  if (is.na(x)) return (NA)
  poisson.test (x, r = 100)$p.value 
}
# test it again
MyPois(hospitals[1,1])
MyPois(hospitals[1,2])
MyPois(hospitals[2,1])
```

## Apply it to the dataset

```{r} 
hosp_p_values <- apply (hospitals, 1:2, MyPois)
hosp_p_values
```

## Identify True/False for the test

```{r}
hosp_diff <- hosp_p_values < 0.05
hosp_diff
```

## This has pulled together
- Knowledge of
    + object types
    + indices and subsetting
    + how functions work
    
## Function gotchas
* Objects don't print unless you use `print` function
* The last value is returned (by default)
* Function may apply to a vector or a scalar
* Changing object types cause problems
* more advanced, environment function created can matter