IPD

1.
Getting Started
with R and Data Science

Data Science Goals,

Vectors,
Functions,
Univariate Analysis (Vector Analysis)

1
 IPD

What does R (Data Science) have to do with

business management or economics?
Or: Why do we have to learn this?
 IPD

Raw Data

This is a part of a company's sales data. The data holds 49 variables (columns) and almost 50.000 observations (rows).
Raw Data – like the above - typically do not directly support data-based decision making or data-driven management.

The goal of data science is to draw meaningful information out of raw data; information that supports decision-making.
 IPD

Data Science with R

What types of meaningful information can we get out of raw data?

 IPD

We use R
to get meaningful information
out of raw data.
 IPD

Why not Excel?

6
[email protected]
 IPD

Vectors

Name an important characteristic of R Vectors:

Working with R Vectors:

• assigning values to a vector (creating a vector):
• example for an element-wise operation on a vector:
• get value at a certain position:
• logical subsetting (element-wise): # return the value if the condition at the position is TRUE
vector_name[condition] # element-wise execution
# also uses [] – position brackets

7
 IPD

Important R Data Objects

Dimension One Data Type Multiple Data Types

1 Vector

R language: vector
DataScience language: variable
Tabular Representation: column

8
 IPD

Programming Concepts

What programming components can you find in the cartoon?

Source; Jeremy Keeshin, ReadWriteCode, p.105 9

 IPD

Programming Concepts

Input is: 4,5,6,7,8,9

Write the R command(s) to get the X * 2 output:

Source; Jeremy Keeshin, ReadWriteCode, p.105 10

 IPD
Data Objects and Functions

Data Objects Code Objects

e.g. Vector Functions

store data elements (values) store instructions, operate on data

functionName() ,
functionName(argument1=value,argument2=value,
vector_name
argument3=value, …)

die_vec c(1,2,3,4,5,6)

11
 IPD

Getting Meaningful Information out of Raw Data

"week1",605,1572,1454,2103,3579,3135,3971,"week2",2470,2062,1884,1786,3694,3258,2690,"week3
",1996,2676,1565,1551,3497,3498,3652,"week4",2999,2749,2522,2745,3284,3343,3142,"week5",1987
,2662,1584,2267,3546,3641,3539,"week6",1275,2353,3287,3388,3753,3936,3173,"week7",2235,2430,
3333,2101,3825,4578,3879,"week8",2926,1941,3353,2906,4018,3714,3656,"week9",2634,3163,3551,2
821,4304,4179,4237,"week10",3249,3585,2066,3427,3765,3691,4259,"week11",2191,3771,3254,3447,
3524,4392,4335,"week12",1650,1863,3491,2738,4638,4432,4183,"week13",4050,3883,2915,2287,415
7,4297,4757

Here is some raw data – daily sales revenues – from a small business.
We start with a univariate analysis, that means we only look at one variable - the raw sales data.

What type of meaningful information can we get out of this variable?

12
 IPD

Univariat Summary Statistics with R

Step 1: Create the vector in R

- how many elements does the vector have?

- what vector type is it?
- what does the summary() command return?

13
 IPD

Univariat Summary Statistics with R

Step 2: Convert the character vector into a numerical vector

Statistical summary functions cannot be applied to a character vector.

→ we need to convert our character vector into a numerical vector

sales_rev_num <- as.numeric(sales_rev)

- does the command work?

- Why do we get a warning?
- What output do we get?

14
 IPD

NA Values - Missing Values

• Week information was replaced by NAs
• NA is not a character value. (It does not have quotation marks.)
• In R, NA is a reserved keyword to indicate that a value is missing.
• When trying to convert a character value into a numeric value, R turns the character value into a NA.
(The character value is lost!)
• The resulting vector contains numerical values and NAs (missing values)

Some functions in R work fine with NA values, others do not work with NA values!
Some functions in R ignore rows with NA values!

mean(sales_rev_num)
median(sales_rev_num)
Output?

Before you start analyzing data, ALWAYS check if there are NA values.
15
 IPD

Univariate Summary Statistics with R

Step 3: We check vector type and length again and run the summary statistics:

typeof(sales_rev_num)
str(sales_rev_num)
summary(sales_rev_num)
length(sales_rev_num)

What does the summary() function return?

16
 IPD

Univariate Summary Statistics with R

Median > Mean

How to interpret this?

- in terms of skewdness?

- in terms of sales data:

17
 IPD
NAs - Missing Values

Before you start analyzing data, ALWAYS check if there are NA values.
Important functions:

• anyNA(x) #Checks, if a vector has ANY NA values

anyNA(sales_rev_num) #Result is TRUE if there is at least one NA value
[1] TRUE

• is.na(x) # Returns a logical vector (TRUE / FALSE)

# TRUE ≡ NA, FALSE ≡ value
is.na(sales_rev_num)
!is.na(sales_rev_num)

• Return the number of NAs / not NA elements

sum(is.na(sales_rev_num))
sum(!is.na(sales_rev_num))

The is.na() function can be used for logical subsetting of the vector.

18
 IPD
NAs - Missing Values

Step 4: We remove the NAs out of our sales data vector

• We use the function is.na() and logical subsetting to achieve this

logical subsetting (element-wise): return the value if the condition at the position is TRUE

• vector_name[condition] #if condition is true, value at the position is returned

• Write the subsetting command that removes the NA values out of the vector.

• How many elements are left in the vector?

19
 IPD
R Script
We save all steps in a script.
A script is a sequence of code lines that can be saved as .R file.

Step 1: Create the vector in R

Step 2: Convert the character vector into a numerical vector
Step 3: We check vector type and length and run the summary statistics
Step 4: We remove the NAs out of our numerical sales data vector
Result: We have a clean numerical vector that can easily be analyzed.

• We comment our code.

• Test the script: Run it and check that vectors are correctly created and shown in environment.
• Save your script as .R file.

Always save your scripts.

20
 IPD

Creating a 2nd Vector - Subtracting Taxes

• In Georgia there is a 20% income tax.

We create a 2nd vector out of the 1st that calculates for each day the sales revenue after tax deduction.

- write the R command:

- use the round() function to round the vector values to full integers.

Getting help in R:
?round()

21
 IPD
Summary

Summarize in your own words your key take-aways from the lesson.
Write it somewhere and copy to chat in a minute.

22
 IPD
Summary

Write down the R functions that you learned today

23