Data Analysis with R

Ways to find help within R itself. Also some valuable internet resources:

stats.stackexchange.com

r-bloggers.com

Most used R data type is numeric. For this reason we are starting with numeric data type first.

Sometimes issues with vectors versus simple numbers might arise. But remember - simple numbers are also vectors of length one! Thus treat all numeric variables same way - vectors of different length.

Very important aspect is vectorization - exploit vectorized operations.

Characters are text variables. Every character variable is also vector of some length.

However, there are different functions to work with characters compared to numeric variables.

Logical variables are also vectors of some length. Values are either TRUE or FALSE.

Note that TRUE/FALSE could be encoded as T/F or even 1/0. Feel free to use either of these, because they are equivalent.

Similarly to other programming languages logical operations are available with logical variables - AND, OR, !(negation). After understanding these in one programming language you will be able to use them in others as well : )

First assignment in this section. Feel free to go back to previous lectures if need quick reminder of some topics : )

Very important data type in R. Think of it as spreadsheets. Multiple rows with information in columns. Each column can be of different type.

Pay attention to how data frame is constructed from vector variables. Each column can be of different data type. Compare it with vectors, where each element must be of same type.

List are similar to data frames.

However, lists are more flexible data type compared to data frames. But flexibility comes with its price also. Usually lists are harder to manipulate.

In second assignment data frames with lists are covered. Also combining numeric and character vectors.

You might wish to stop here and spend some more time practicing. Very important to understand data types well - this is main block of language. They will be used in all further sections : )

Package is simply bunch of functions designed by professionals to perform specific tasks.

R has > 5000 packages available for free. Whatever issue you will have - most likely there will be package to solve it : )

Multiple data importing options available. However, in this video we will cover mostly used ones:

1. Text
2. Previously saved variables in .Rdata files
3. Excel
4. "datasets" package

If data could be imported, it can be exported as well.

By doing data analysis you will most likely need to save some output. Data export methods covered in this lecture includes:

1. Text

1. .Rdata files
2. Excel

Here we will combine import, some manipulation and export together. This is common workflow for data analysis.

In latter lectures we will learn more advanced manipulation techniques : )

Very often some logical conditions should be checked before deciding which actions should be taken.

If-else statement usually looks like:

if (condition 1) {

Insert code to execute

} else if (condition 2) {

If condition 1 is not satisfied and condition 2 is - execute this part of code

} else {

If none of condition 1 or condition 2 are satisfied - execute this part of code

}

While loop is used when we need to perform specific code as much times as given condition is satisfied. Stop executing code once condition is not satisfied.

While structure looks like:

while (condition) {

# Code to be executed

}

Make sure condition will be satisfied finally to avoid infinite loop : )

For loop is used when we have bunch of elements and would like to perform some code with each of them. For example run through list elements and print particular element of it.

Note that If-Else or while statements could be used inside For loop. Opposite holds true as well.

Structure of for loop:

for (index in set_of_elements) {

# Perform some manipulation with "index"

}

Checking your knowledge about if's, while and for loops.

Be sure to try creating combinations of these. Any control within each other. Try combining them to accomplish some goal - try to think about something(even not very realistic in real world). Most important to understand logic and syntax for this section.

Function is chunk of code to perform specific task. You provide multiple arguments to function and get the result back. Results are returned using return(results) statement. For example:

FunctionName = function(argument1, argument2) {

# Code performing some manipulation

return("include results here")

}

Once function is sourced, variable of type function will be created and you will be able to call it with necessary arguments.

Usually functions are more complicated. This lecture demonstrates just that.

Combining functions with while/for loops and if's is very good exercise!

R is functional programming language. So functions is very often used for almost all jobs you need to do.

Either you can use built in functions(like mean(), read.csv()) or packages(XLConnect for example) or even write your own functions.

Experiment with functions. Try to call one function from another function. Think about ways to exploit functions for your needs : )

More practice never hurts!

In this section basic graphics will be learnt. Starting with line and scatter charts.

We will be using "base" R graphical engine. However, there are more graphical packages available:

• ggplot2 - http://docs.ggplot2.org/current/ and http://cran.r-project.org/web/packages/ggplot2/index.html - Currently most popular graphical package in R.
• lattice - http://cran.r-project.org/web/packages/lattice/index.html

However, these packages are more advanced and takes some time to learn them. Firstly, make sure you are comfortable comfortable with "base" graphs and then try other capabilities as well.

Bar, Pie charts covered here. More manipulation options added. Note that usually same arguments for manipulating plots are used in all chart types - Bar, Pie, Line, Scatter.

Best way to learn graphics is to firstly have a problem and then figuring out how to plot results.

After completing assignment think about data you want to visualize for some time yourself and do it right now!

This will require a lot of topics covered in this course - acquiring data, importing it, manipulating, plotting.

Named vectors are just simple vectors, but with added names to it. Same calculations applies as to simple vectors.

You can convert named vector to simple vector and back at any time.

Matrices is similar to data frames, because both have 2 dimensions. However, all columns should be of the same type in matrices.

Assignment of names to rows and columns possible.

Mathematical calculations might be performed. See http://en.wikipedia.org/wiki/Matrix_(mathematics) for more.

Dates and times are crucial component when dealing with data generated during time.

A lot of date/time formats exists in computer science. They might be tricky to use at first. Only practice helps here : )

Using R famous apply() function for data frames. This function summarizes data frame by rows or columns.

Possibility to use custom(your own) functions together with apply().

Some additional reading - http://www.r-bloggers.com/using-apply-sapply-lapply-in-r/

Similary to apply for data frames, lapply or sapply used for lists.

Here we don't have rows/columns parameter. lapply/sapply runs through all elements of the list applying your provided function.

lapply will always output list, while sapply will try to simplify results to vector or matrix.

Time to complete assignment for both apply and lapply/sapply.

Cheers! : )

Super useful package reshape2 written by Hadley Wickham - thanks to him.

This allows to structure any given data frame in totally new form of data frame suitable to your needs.

Don't forget to check out http://cran.r-project.org/web/packages/reshape2/reshape2.pdf for some good examples.

Good time to reshape data yourself!

Yet another useful package from Hadley Wickham.

Package allows to take one data format(list, data frame, matrix), slice depending on selected columns and finally combine result back to new format(list, data frame, matrix).

Check your knowledge about plyr package. Let me know if any questions arises.

In this section you will conduct data science project yourself!

Find your own data or use suggested one - check next lecture. No strict guides how to do this task, but there are some suggestions available in next lecture if needed.

Don't be afraid to experiment. Raise hypothesis questioning it with your data.

Finally let me know how you are doing and feel free to share your work. I would love to check it out : )

Cheers!

Possible project description. Feel free to adjust it if needed to meet your needs.