Logical conditions can become quite complex and it could be easier to create a function to deal with them.

greaterThan <- function(x, than) { x > than } evenNumber <- function(x) { x %% 2 == 0 } ofLenght <- function(x, l) { nchar(x) == l }

Once defined you can use those functions like you did for bare conditions. The way it works is the function will be applied to every element of the vector or list.

Get only even numbers of the list:

a[evenNumber(a)]

Get elements greater than 4:

A[greaterThan(A, 4)]

Get only strings of the length 2:

b[ofLenght(b, 2)]

Instead of [] parenthesis, one could use dedicated functions: Filter or subset. They may not seem much at the moment but they bring value once working with data frames instead of simple vectors and lists.

Filter function takes 2 parameters - condition function and data container. Condition function needs to take 1 argument, which will be mapped for every element of the vector or list.

Filter(evenNumber, a)

To use greaterThan function (that takes also than as an argument) to limit result to only elements greater than 4 one could use inline function syntax:

Filter(function(x) { greaterThan(x, 4) }, a)

Another useful function is subset. For vectors and lists it works very similarly to the [] filtering.

Using a bare condition:

subset(a, a > 4)

Using functions in subset:

subset(A, greaterThan(A, 4)) subset(b, ofLenght(b, 2))

One of the ways of choosing the data is indexing. You can do it in two ways. One is to specify the position of the data, second is to provide a logical value.

If you know the numbers of the elements you can easily provide the list and filter by it. Let's say we want 1st, 4th and 7th element.

indexes <- c(1, 4, 7) a[indexes]

To get the rest of the elements you can use - operator.

a[-indexes]

If you already have the logical values for the elements you want to include or exclude, you can use it to filer also. This is consistent with how applying logical conditions works.

logical <- c(T, T, F, T, T, F, F, T, F, F) a[logical]

Analogically to get the excluded elements you can use - operator.

a[-logical]

Filtering data becomes much more interesting when working with complex types, especially data frames.

Let us create an example data frame. You can imagine it contains poll data gathering information of age, gender and if the person is a smoker.

population <- data.frame( age=c(20, 15, 31, 45, 17, 27, 35, 41, 52, 23), gender=factor(c('F', 'F', 'M', 'M', 'F', 'M', 'F', 'F', 'M', 'M')), smoker=c(T, T, F, T, F, F, T, F, T, T)) population

Every column is accessible using $ symbol.

population$age population$gender population$smoker

To filter data by column information we can use [] similarly to how we did this for simple types. There is a tweak though, column information needs to come after the ,.

Instead of $ symbol you can choose to select the column by the number.

population[, 2] population[, 3]

Or a list of numbers.

population[, c(2, 3)]

You can even change the order of columns this way.

population[, c(1, 3, 2)]

If you don't know the order of columns you can use names.

population[, c('age', 'smoker')]

When working with data frames filtering is usually applied to the samples. R platform is providing several interesting techniques.

First, if you treat columns of the data frames as vectors, you can easily filter them.

Get ages of the males in the population:

population$age[population$gender == 'M']

Get information about smoking for people over 30:

population$smoker[population$age > 30]

You can also put together both conditions:

population$smoker[population$gender == 'M' & population$age > 30]

As you can see the output is just the list of values of the corresponding column, without any other information of samples. To get them as the result, the filtering could be dome using [] and the ,. Similarly to filtering by columns, this time the condition should be before the comma.

You can specify which rows you want to choose. For example get rows of indexes from 2 to 5 and the 8th one:

population[c(2:5, 8), ]

If you want to apply the condition, do it analogically to the vector/lists mechanisms. Following up the examples above we get rows for males, people over 30 and with the combined conditions.

population[population$gender == 'M', ] population[population$age > 30, ] population[population$gender == 'M' & population$age > 30, ]

You can achieve the same results using functions instead of bare conditions.

ageGreaterThan <- function(x, than) { x$age > than } population[ageGreaterThan(population, 30), ]

Last step has shown that when applying multiple conditions, filtering statements can become quite complex and may not reflect the intentions clearly.

This is where the subset function can help. It allows you to skip constantly writing the name of the data frame by providing it as an argument.

If you want to filter data frame by age you could use the following syntax:

subset(population, age > 30)

You can of course use ageGreaterThan function created in last step, but it doesn't seem to improve the readability of the statement.

subset(population, ageGreaterThan(population, 30))

You can see subset function being useful when putting conditions together. For example the following statement is much easier to read when getting the population of males over 30:

subset(population, gender == 'M' & age > 30)

Apart from filtering samples you could use the function when picking up specific columns. Nothing should stop you from putting both filters together.

If you want to know only the gender and smoking information for people over 30 you could choose second and third column only.

subset(population, age > 30, select=c(2,3))

Or you can work with column names:

subset(population, age > 30, select=c(gender,smoker))

Note that this time column names don't have to come in as char types. If you want you could also use strings.

subset(population, age > 30, select=c('gender','smoker'))

Another simple way of removing the complexity of statements is to use attach and detach functions. They 'load' and 'unload' your data frame and give you the access to the columns directly. In our examples you could use the following code:

attach(population) gender population[age > 30,] detach(population)