Graph Types

But…why?

Number of emergency department visits due to energy drink consumption.

Number of days energy drinks are consumed (in last 30 days) and score on Perceived Stress Scale.

Emergency department visits due to adverse reactions from or abuse/misuse of energy drinks.

Make sure you're data is “tidy”:

• Observations are rows.
• Variables are columns.
• First row (and only first row) consists of variable names.
• No extra stuff outside the rectangle.

• Categorical (nominal, qualitative, factor)

Classifies data by category.

color, species, sex

• Quantitative (numeric, scale, interval/ratio)

Numerical measurements, usually with meaningful units.

height, GDP, score

CAREFUL!

Numbers are not always quantitative.

Do you own a car?

0 = “No”, 1 = “Yes”

What is your zip code?

Risk factors associated with low infant birth weight, collected at Baystate Medical Center, Springfield, Mass during 1986.

189 women measured on 10 variables.

• Single variables usually won't answer very interesting questions by themselves.
• Graphs of single variables are often valuable for exploring your data, but generally not suitable for inclusion in the final product.

Mother's race (white, black, or other)

# A tibble: 189 x 1
   race 
   <fct>
 1 Black
 2 Other
 3 White
 4 White
 5 White
 6 Other
 7 White
 8 Other
 9 White
10 White
# ... with 179 more rows

• Frequency table

• Relative frequency table

Click here to see how to improve a pie chart.

• Why do pie charts suck?

“Pie Charts Are the Worst”

“Countdown of Top 10 Reasons to Never Ever Use a Pie Chart”

“Save the Pies for Dessert”

• And to be fair, one guy who defends pie charts:

“Why Tufte is Flat-Out Wrong about Pie Charts”

Infant birth weight in grams

# A tibble: 189 x 1
     bwt
   <int>
 1  2523
 2  2551
 3  2557
 4  2594
 5  2600
 6  2622
 7  2637
 8  2637
 9  2663
10  2665
# ... with 179 more rows

• Tabular summaries

(“Five-number summary” or other quantiles)

• A bunch of other types I don't prefer:

boxplot, stem-and-leaf plot, dotplot

There are at least six elements of a plot that can be assigned to variables:

• x-axis (horizontal axis)
• y-axis (vertical axis)
• facets
• color/fill
• size
• shape (e.g., dots vs crosses, solid vs dashed lines, etc.)

Mother's race and the number of previous premature labors.

# A tibble: 189 x 2
   race  ptl  
   <fct> <fct>
 1 Black 0    
 2 Other 0    
 3 White 0    
 4 White 0    
 5 White 0    
 6 Other 0    
 7 White 0    
 8 Other 0    
 9 White 0    
10 White 0    
# ... with 179 more rows

Mother's race and infant birth weight in grams

# A tibble: 189 x 2
   race    bwt
   <fct> <int>
 1 Black  2523
 2 Other  2551
 3 White  2557
 4 White  2594
 5 White  2600
 6 Other  2622
 7 White  2637
 8 Other  2637
 9 White  2663
10 White  2665
# ... with 179 more rows

Mother's age and infant birth weight

# A tibble: 189 x 2
     age   bwt
   <int> <int>
 1    19  2523
 2    33  2551
 3    20  2557
 4    21  2594
 5    18  2600
 6    21  2622
 7    22  2637
 8    17  2637
 9    29  2663
10    26  2665
# ... with 179 more rows