Data Mining

Question 1

What 4 major issues do you need to addresss when clustering?

Answer 1

1) Similarity / dissimilarity measure
2) Standardization/normalization of numeric variables
3) Re-coding categorical variables
4) Number of clusters

Question 2

Describe ways you might improve your model?

Answer 2

1) Divide and conquer: multiple models for different areas of data
2) Derive new, delete, merge variables
3) Impute missing values
4) Normalize/standardize numeric variables; remove outliers