Stage 3: Data analysis: measurement of complexity

This stage involves extraction and analysis of the appropriate data from the constructed CAD models. The results are applied to the formal model developed in Stage 1 in order to give us measures of the complexity of the models.

Included in this stage is the development of appropriate data extraction and comparison techniques. Among the criteria used in comparison are:

To provide an illustration of issues and what will be involved in this analysis, a group of AutoCAD R14 files produced by a few students are summarised in the table below. For a detailed look at the models, click on the images to open a new window (requires WHIP! plugin or comparable for viewing DWF files, Cosmo viewer plugin or comparable for viewing VRML files).

File ID.

Project ID-
Student ID-
Model seq. no.

Description of file and counted element types
File size Objects Layers Block definitions



2D drawing of Magney House: architect Glen Murcutt
415K 6975 13 2



3D model of Magney House: architect Glen Murcutt
Produced by the author of File 1
3067K 693 9 1



3D model of Magney House: architect Glen Murcutt
929K 934 18 19



3D model of Short House: architect Glen Murcutt
2936K 3031 70 25



3D model of Short House: architect Glen Murcutt
Includes part completed 2D drawing and text
433 (initial)

Quantities of objects, layers and blocks shown above are as calculated by AutoCAD when using the commands STATUS, LAYER and BLOCK. The term object includes graphical objects such as arcs and polylines, nongraphical objects such as layers and linetypes, and block definitions. The figure for blocks includes both user blocks and unnamed blocks.

The selected files are drawings and models produced by students in an introductory one semester AutoCAD course. As might be expected from non-expert users, the files vary substantially in their respective quantities of objects, suggesting different modelling approaches among files, and even inconsistencies within files. Files produced by more experienced individuals are likely to show more consistency.

The main observation that can be made from this small sample is that measured quantities of various file elements are not especially revealing or conclusive on their own, but may be useful indicators of where one should look for determining certain characteristics of a file, such as inconsistencies of technique, or extensive use, or non-use of various methods of organisation or production. Individual points worth noting are that:

Additional analysis of this small but diverse group of files will be undertaken to further establish the extent and impacts of their differences.