Uncertainty in Geospatial Data
Download as PPTX, PDF1 like148 views
This document discusses various measures of uncertainty in geospatial data derived from confusion matrices. It introduces the concepts of omission and commission error, which measure errors of exclusion and inclusion, respectively. It also covers total accuracy, limitations thereof, and Cohen's Kappa coefficient as alternative measures that account for agreement by chance. The document aims to help learn about uncertainty factors in raster data modeling.
![COHEN’ KAPPA
• Formulas: [(a + d)- (((a + c)(a + b) + (b + d)*(c + d))/N)]/[N - (((a + c)(a + b) + (b + d)(c
+ d))/N]](https://image.slidesharecdn.com/uncertainty-161217141655/85/Uncertainty-in-Geospatial-Data-7-320.jpg)
Uncertainty in Geospatial Data
- 1. UNCERTAINTY IN GEOSPATIAL DATA By : Ehsan Hamzei (810392121)
- 2. GOAL • We want to Learn about Uncertainty Factor in Raster Model. • All of them derived from “Confusion Matrix” (or Ambiguity Matrix).
- 4. OMISSION - COMMISSION • Error of omission ( Producer’ s acc.) : • proportion of values in reality, which were interpreted as something else: Sum of column’s non-diagonal elements / column total (For Corn 8/20) • Error of commission (User’s acc.) : • proportion of values which were in reality found to belong to another class: Sum of row’s non-diagonal elements / row total (For Corn 6/18)
- 5. TOTAL ACCURACY • Also called: Correct Classification Rate! • Formulas: TA = (a + d)/N • Where N = a + b + c + d
- 6. LIMITATION OF TOTAL ACCURACY • Consider a 2-class problem • Number of Class 0 examples = 9990 • Number of Class 1 examples = 10 • If model predicts everything to be class 0, accuracy is 9990/10000 = 99.9 % • Accuracy is misleading because model does not detect any class 1 example
- 7. COHEN’ KAPPA • Formulas: [(a + d)- (((a + c)(a + b) + (b + d)*(c + d))/N)]/[N - (((a + c)(a + b) + (b + d)(c + d))/N]
- 10. THANKS…