Community Data Science Course (Spring 2017)/Day 7 Notes

A bit of vocabulary: types of data

• Categorical: data that takes a qualitative value. For instance, city, weather condition (icy, rainy, sunny).
• Ordinal: data that can be used to order things. Generally, the important thing about ordinal data is that 'scale' isn't important.
• Cardinal: data that counts something. Examples: age, temperature, GDP.

Do this: Find two examples of each kind of variable in the SDOT crash data.

Plotting:

1. Pick your x-axis variable.
2. Pick your plot type.
   1. 'cardinal' and 'cardinal': Scatter
   2. 'categorical' and 'cardinal': Bar or histogram.
   3. 'categorical' and 'categorical' and 'cardinal': We'll see some examples of this in python, but a pivot table with color is often useful.
   4. 'ordinal' and 'cardinal': Bar, line chart, or histogram.
3. Pick your y-axis variable.

Do this: Make the following plots (for each figure out which kind)

1. Count accidents by road conditions
2. Build a plot of number of people involved and injury count.
3. Build a plot of accidents by date.
4. Use a stacked plot to see if speeding is more associated with some weather conditions.

General Advice

• Don't overthink it! Excel has dozens of fancy plots, and there is a good reason to use all them. However, don't introduce a more complicated plot where a simple one will do.