What is experimental computer science; Computing science research methodology; Statistics of interest; Exploratory data analysis; Hypothesis testing; Sampling; Experiment design; Explaining performance; Parametric modeling, regression; Generalization.

Course Objectives

This course gives a rigorous background in empirical methods for students working in any area of applied computer science. It covers experimental design, probabilistic modeling, exploratory data analysis, hypothesis testing, and system tuning. It encourages the use of computer-intensive statistical methods as an alternative to classical statistics.

Recommended or Required Reading

Cohen, P. R. (1995). Empirical methods for artificial intelligence, MIT Press.

Knuth, D. E. (1997). Seminumerical Algorithms. Reading, MA, USA, Addison-Wesley.

Learning Outcomes

1. Understand the importance of empirical research.

2.Construct a causal model between input and output variables.

3.Apply hypothesis testing.

4.Adopt computer-intensive statistical methods.

5.Assess performance.

Grading

Midterm 30%

Project 30%

Project Publication 40%