Beyond Accuracy and Precision
We discussed the issues of accuracy and precision in last lecture, but we should also look a bit deeper at error and other related issues defining data quality.
We are now well versed on how important “clean” spatial data is for mapping and analysis, but we may not know how good the data we download may be. We will go through a couple of excellent pages that describe these issues.
Foote and Huebner – 1995
Lets look a a web page written by Kenneth E. Foote and Donald J. Huebner (the Geographer’s Craft Project, Department of Geography, The University of Colorado at Boulder – 1995). This is an old document, but still relevant today.
The page is no longer held at University of Colorado, but Penn State has carried on with the content – Penn State Foote and Huebner
Another great presentation using much of the Foote and Huebner document by Austin Troy
There are many things to consider from these pages, but lets look at portions of the information that we should consider in this class and for your projects.
- Spatial accuracy and Precision (we discussed this last lecture)
- Attribute Accuracy and Precision (we discussed last week as well)
– think of examples (i.e is rounding effecting accuracy or precision)
Troy slide 12 Foote 4.2.3, 4.3.3
– can we present data with inaccurate results using attributes
Troy slide 13 Foote 4.3.3
- Conceptual and Logical errors
– Conceptual – catagorizing your data to meet the purpose of study Foote 3.3
– Logical – how do you put your data layers together Foote 3.4
- Sources of data inaccuracies of imprecision Foote section 4
– Age of data
– Area completeness
– Scale and resolution
– Processing errors – digitizing and overlays (slivers – loops)
- Propagation and Cascading Foote section 5
– Propagation, an error in one layer adds to errors if the layer with the error is used in another layer
– Cascading – propagation running amok – see Troy slide 17
– VERY important
– When you download data, pay attention to other pieces of information describing aspects of accuracy/precision as well as necessary info such as projection
– Data about data, helps in deciding how or whether to use the layer
– We will review this in tutorial while exploring data sources