| Penn State METEO 474: Syllabus |
| On this page: | Textbooks and Web Resources | Grading | Course Contents |
| Meteo 474: | Home Page | Lesson and Assignment Due Dates | |
OBJECTIVES
- Learn those computer methods needed for statistical analysis and forecasting of the weather.
- Learn to apply those methods to develop accurate and robust weather forecast systems.
APPROACH
Ongoing advances in statistical forecasting techniques are making automated weather prediction systems ever more competitive against the human forecaster. At the heart of this revolution is the advance from simple linear models of cause and effect to complex nonlinear models mimicking the way forecasters think. These artificial intelligence systems have proven able to handle forecast problems involving multiple possible mechanisms for the same outcome, long the forte of the human forecaster. With the forecaster's advantages shrinking, the speed of automated systems is becoming a telling factor in the weather forecast market. To succeed in this changing environment the operational meteorologist must become a builder of forecast systems, one who uses meteorological insight and statistical knowledge to create ever more accurate forecast systems. Meteorology 474 addresses these issues, providing hands-on experience building and testing the forecast systems of the future.
The course's focus is on learning to use state-of-the-art computer programs (e.g. WEKA) to develop advanced statistical weather prediction system. The class will inclujde lecture periods to learn what statistical methods work and why, demonstration/experimentation periods to gain hands-on experience with these methods, and laboratory periods to develop your own forecast systems using them. The laboratory assignments will use real-life weather forecasting problems, generally involving high impact weather such as thunderstorms or flight-limiting fog. The final assignment will involve a head-to-head competition (i.e. a forecast contest) between statistical forecast systems of your own design. Examinations will test how well you understand the strengths and limitations of the statistical methods, how to get the best out of them, and how to assess their accuracy.
TIME and LOCATION
Class times and places:
TR 9:45 - 11:00 in 003 Walker
Office hours: TWR 11:00 till 12:00 in 621 Walker
Textbook (Required)-- "Data Mining, Practical machine learning tools and techniques with Java implementations" by Ian H. Witten and Eibe Frank
Web Resources for WEKA
- Overview - http://www.cs.waikato.ac.nz/~ml/
- Download - http://www.cs.waikato.ac.nz/~ml/weka/index.html
- Tutorial - http://www.cs.waikato.ac.nz/~ml/weka/Tutorial.pdf
- File Format - http://www.cs.waikato.ac.nz/~ml/weka/arff.html
Course grade is based on successful completion of four laboratory/homework assignments and two examinations.
Academic integrity: You are required to turn in your own team's work for each assignment. While it is expected that your group will freely exchange advice and assistance with other teams, the work you turn in must be your own team's. Programs copied from others, including persons not taking the course, will be given a grade of zero. If, however, you find a public domain utility on the web that you feel it would be to your advantage to use, you may ask for my permission to incorporate it (with appropirate credit given) into your project. I will decide whether to grant such advance permission on a case-by-case basis based on whether the code in question is a general purpose utility freely available to the global public.
For more details see the lesson and assignment list.
The Meteo 474 Web Pages are maintained by Dr. George Young.
This page was last modified on March 25, 2003.