Event 07: Classification

You are given the Population Density [spreadsheet format] figures for Africa (in the year between 1999 and 2000?)...

For each method: figure the "class limits" (range of values assigned to each class) with a few (conflicting?) rules in mind.

• no class should overlap with any other
• the whole range of the attribute should be included
• resolution of the attribute should be "reasonable" (consider the significant digits)

How to do each method:

Quantiles: generate an "equal" number of countries per class.

Classify the data using Quantiles to create five classes. First divide the number of countries by the number of classes, giving a whole number and remainder. Allocate the remainder in some even manner so that classes only have two possible whole numbers of countries. Use the sorted data table you are provided to count off the appropriate number. These are the countries that belong in each class. Class ranges are expressed as exhaustive, non-overlapping values to some precision i.e 101 - 200; 201-300, etc. or 1.00 ­1.99, 2.00- 3.99, etc. Generate class ranges that suit this attribute.

Equal Interval:

The following formulas are from Dent page 145. Note that there is a typographical error in the text for the formula at step 2. The numerator should be R not P.
Calculate the range of the data values.
R = H - L
R = Range; H = Highest data value; L= Lowest data value
Calculate the common difference (or class width).
CD = R/NC
CD = Common difference (class width); NC = Number of classes
Calculate the class upper limits:
Low = L + 1*CD
Medium Low = L + 2*CD
Medium = L + 3*CD
Med High = L + 4*CD

Natural Breaks:

Observe the breaks on the Chart side, observe the "differences" column.

Pick four breaks that give you a good rendition of the overall distribution, the absolute value of the difference may be less important than the relative difference (proportional to the value?). Consider how many countries are in each class, but this is NOT quantiles...

Answer sheet:
NAME:

Section:

QUESTION 1: What level of measurement best describes the data in the NAME field? What level of measurement best describes the data in the POP_DENSITY field?

For each classification method fill in the following tables:

Quantiles

	number of countries	Class limits (for a legend)
Low
Med. Low
Medium
Med. High
High

Equal Interval

	number of countries	Class limits (for a legend)
Low
Med. Low
Medium
Med. High
High

Natural Breaks

	number of countries	Class limits (for a legend)
Low
Med. Low
Medium
Med. High
High