Translations:Basic Excel/147/en
We recommend the following approach for working with a new dataset:
- Save a copy of the original data.
- Never make changes to the original dataset.
- Is there data on more than one worksheet in the workbook?
- Inspect each worksheet.
- Do not assume that an unnamed worksheet will be empty.
- Make sure you understand what you have been asked to do.
- What data will be needed to complete this task?
- Must all workbook data be processed?
- Does this workbook represent the complete task, or will additional (similar) workbooks be arriving?
- Outline the structure of the final structure that will be needed to complete the task.
- List the headers and data categories.
- Make a note of the problems or issues that make the original dataset unsuitable for immediate analysis, or that must be changed so that it fits your proposed structure. Look for:
- dataset not being contiguous (blank rows or columns separating components of the dataset)
- repeated heading row
- inappropriately formatted header row
- multiple header rows
- merged cells in header rows
- merged cells in columns
- repeated columns, or columns containing data that is not needed for the analyses
- this material does not need processing
- multiple pieces of information stored in a single column
- categorical data with inconsistent categories
- problem data
- some incorrect data will never be identified
- some incorrect data will only become obvious once the data summarisation has been completed (this will be the next exercise).