The post-processed data is useful because each record contains a large quantity of information, including the entire command line. Previous studies have recorded command frequency patterns, but have not recorded entire command lines. In addition, the formatted post-processed data will make future analysis relatively easy.
Analysis results will provide useful data for designers of future command interfaces, and they may suggest improvements to the Unix C Shell. The analyses completed to date have shown that a few commands account for the vast majority of the data. In most cases, usage patterns are relatively simple. The average user only uses about one-fifth of all commands, and the vast majority of all command lines contain only one command. Users, however, often alter their environment by using aliases.
In many cases, analysis produced predictable results. Users often execute commands in the background that use a relatively large amount of CPU time. Certain complex features are very seldom used. Programmers and graduate students tend to have more complex usage patterns than secretaries.
One surprising result was that almost 30 per cent of all the data was generated by system command files. The daemon program generated 28.4% of all commands. This result, however, is probably a unique characteristic of the machine used for data collection, and is not characteristic of typical C Shell use.