Thesis Topic: Outlier detection in Skew Distributions.
Brief Description: Standard techniques for outlier detection do not work very well in skew distribtions. In this thesis we will develop some techniques which work better in this case.
For full description, see my homepage.
I have given you feedback on Chapter 1: Please incorporate and re-write.
OUTLINE:
1.1 INTRO: Methods of Data Summarization - History
Good Data summary depends on data -
Normal Data: Mean and SD
UNIMODAL and SYMMETRIC: Boxplots
UNIMODAL and SKEW: IHA (your) method
Motivate and Justify. Prove better for theoretical dist and also for real data sets.
Iftikhar: I would like you to read these books (the ones available and choose nice examples of data analysis with real data from them to incorporate into the textbooks we are working on -- that is, Intro to Statistical Analysis and ALSO, Intro to Applied Econometrics.
The Elements of Graphing Data, William Cleveland, Hobart Press
Visualizing Data, William Cleveland, Hobart Press
The Visual Display of Quantitative Information, Edward Tufte, Graphics Press
The Grammar of Graphics, Leland Wilkinson, Springer
ggplot2: Elegant Graphics for Data Analysis, Hadley Wickham, Springer
Lattice: Multivariate Data Visualization with R, Deepayan Sarkar, Springer