You are on page 1of 2

*A VERY* Hypothetical CASE SCENARIO:

Congratulations!!! You are the Team Lead at K&N BA Department.


 You are hoping to expand your services and export poultry to Europe
[BUT: this case is set in *Cold War Europe*]
 You want to identify some clusters of countries in this region [method:
hclust]
 Ideally, you would want to find 3 regions that will be led by your sub-
leads.
 Motivation: Any discoveries you make will likely inform your marketing
strategy

Here is what you need to do:

1. Explore Protein Consumption Dataset


 What do the numerical values represent?
 How many countries are being represented?
2. Prepare Data:
 Since our goal is hclust [distance = Euclidean] what are the only kinds of
variables that we can use?
 Can we just use the numerical features in data as is? Or should they be
transformed in some way? Why?

3. Distance Matrix:
 The input of our distance function [that we saw in Session 15] should be a
matrix

4. Clustering Algorithm:
 Recall Linkage Criteria. Produce results for:
o Average linkage
o Complete linkage

5. Time to interpret:
 CONTEXT time! Think Historical, Political, Geographical
 Which linkage method works better for you?

You might also like