Geographically Weighted Regression: A Brief Overview and Some Recent Developments
Event details
| Date | 07.03.2011 |
| Hour | 16:15 |
| Speaker | Prof. Stewart FOTHERINGHAM; National Centre for Geocomputation, IE |
| Location |
GR B3 30
|
| Category | Conferences - Seminars |
Geographically Weighted Regression (GWR) is a statistical technique developed by the author and two colleagues to examine possible spatial non-stationarity in relationships. The concept is extremely simple but extremely powerful and can easily be extended to domains other than regression and is applicabale to a wide range of areas. The power of GWR is that it replaces a global model with a series of local models each of which reveals processes operating at specific locations. Traditional models tend to produce parameter estimates which are global averages of these local conditions. The basic premise of GWR will be described and two examples shown of its use. Several extensions of this basic framework will then be discussed briefly.
Links
Practical information
- General public
- Free
Contact
- Prof. François Golay, LASIG