Calculating Metrics for Spatial Data Quality

GIS 5935 Special Topics in GIS 
Lab 1

Learning Objectives

Understand the difference between precision and accuracy

Calculate vertical and horizontal position accuracy and precision

Calculate root-mean-square error (RMSE) and cumulative distribution function (CDF)

Calculating Accuracy and Precision 

Accuracy is how close values are to the "truth", assuming that there is one truth that can be observed. Precision on the other hand is how close collected values are to each other. In order to calculate precision the average location of all the points is calculated and mapped. The distance from the average point is then calculated and used to find different significant percentiles. In this case, the 50th, 68th, and 95th percentiles were calculated and represented as a multiple ring buffer (see map below). Accuracy is calculated by determining the distance between the average point and the actual point. In this case, a separate reference shapefile with the "true" location was used as a comparison. 

Results:

Horizontal Accuracy: 3.23 meters

Horizontal Precision(68%): 4.427 meters

Measurements of accuracy and precision are often subjectively defined(Bolstad 7th Edition). Most of this data set appears to be fairly accurate but somewhat imprecise. However, the presence of many outliers around the outer edge of the buffer degrades the data quality. 68% of the data points are within 4.43 meters of the average. This in turn is relatively close to the actual  “true” reference point. The measures of accuracy and precision show that this dataset it more accurate than it is precise.