The Moshi System

The various Enable .dbf files were assembled from various hard and floppy discs.

A program in the Clipper programming language had to be written to process these files, which, although they had a .dbf suffix, were not of dbase format.

The yardage and number sold data was entered manually from the raw data sheets into a separate dbase3+ database.

There was a problem with the actual market-day date within each monthly Enable file - this market-day date was not actually stored. However, Mr Swai, the data input person, assured me that they were always entered in week sequence. Therefore a program was written which read the yardage figures from the yardage database, and which then assumed that the sample size for each week was directly proportional to the yardage (with zero sample if no data of course); this program then wrote the appropriate market-day date against each line item data record in the raw data database. Note however that because of this technique, which for time and location reasons I did NOT verify against the paper records, all the market-day records for Arusha and Moshi must be regarded as slightly suspicious, since maybe on average 5% of the records for each market-day may actually belong to the week before or the week after. I recommend that the master database be checked against the paper records for Moshi and Arusha. This would probably take me 2-4 days to complete, and must be done with the original paper records from the Moshi Office to hand (which probably comprise a 5-ream box weighing 14 kg).