These rules should be followed about getting data for Daisy :-
- Always use the raw data! More...
- Resolve sub records and present the database as a two-dimensional table! More...
- Try and use computer data which has been collected automatically or electronically. Humans are so error prone! More...
- Always work on a copy, as Daisy needs to look at the whole database without parts of it being changed by others. More...
- Microsoft Excel workbooks and worksheets present their own set of cautionary rules. More...
Note that the first rule of always getting raw data is very important.
If You Can't Get Raw Data, Ask Why!
And Forcefully!
Always Use the Raw Data
Why?
- With Daisy you are looking for linkages and nuances between fields and records. Use a simplified form of the database with some records aggregated together and this information may well be lost.
A sad thing is that some systems, such as those used to control petrol stations in the UK, can only produce aggregations. This makes the data useless for analyses such as resource planning.
- The overhead of bringing all fields into Daisy is very small, so simplifying may mean that the field you require to complete the analysis is not there.
As hard discs are now in the Gb range, this is not really the limitation it was when Daisy was first written.
- Raw data often contains serious errors, such as missing fields and incorrect entries.
It would be churlish to suggest that the owners of data apply censorship before delivering it to the analyst!
- In many instances, raw data is produced electronically without any human intervention. So it is nearer the truth!
- Obviously, as you refine the analysis, and need to speed things up, the database can be simplified by removing unwanted fields.
But don't do this, until you prove that they are really unwanted.
If you are doing the research as a consultant to a company or organisation, always insist on the raw data.
What if it all goes wrong, or you find results they dislike. Who gets the blame, if you're working on a subset or simplification of the raw data?
Resolve Sub Records
Why?
- Daisy must work with data in this form, as it adds another dimension of nodes and links above the database.
Remember, that James Miller wrote one of the world's first relational databases and should understand how difficult the concept is to explain! Especially, when you have one-to-many relationships.
- When resolving sub-records, take the advantage of adding other fields that may be useful in the analysis.
For instance, in a pharmaceutical system, where the main database recordset contains say drug codes, with such things as price and drug name in a related table, why not add these to the database to be analysed.
Use Computer Data
Why?
- It's quicker to convert data than input it.
- Years ago, when doing the work for Lloyds Bank, James Miller worked with the late Bob Wesson, who was very much against inputting data. Especially, if you could generate it from a model or find it on a computer somewhere.
Bob's measurement systems had the effect of not disturbing what he was trying to measure, by asking people to fill in forms or enter data.
- Humans are very prone to errors.
Using Daisy to check databases and especially those that have been input manually is essential!
Always Copy the Database
Why?
- Because Daisy needs to have access to all of the database as it draws a chart of everything at a particular point in time.
- Copying the database allows other users to access the system without being affected by the analysis.
Daisy and Excel
Note :-
- Daisy has an intelligent opening dialog that is used to open Excel worksheets.
- Daisy needs to have the field names in the Excel worksheet.
Normally, they are in row 1 with the data starting in row 2. An alternative form uses 9 and 11, but Daisy can also be used to locate them.
- Daisy can read all forms of data from a worksheet.
Numerics, dates and times are converted to text strings.
- Excel sometimes makes the wrong decision about how it stores data.
An example is phone numbers, such as 441440783789, which get stored as numerics rather than text.
A full treatise on Linking Daisy to Excel is given under Techniques or in Section 9 of the Daisy Manual.