Preparing Health Data for Power BI

5.2.1 Why preparation matters

The first two rows use different capitalisation for the same apparent facility name. Check whether they are separate records or whether one is a duplicate. If you import the data without examining it, the dashboard might treat one facility as two categories or count a record twice.

5.2.2 Start with a well-structured dataset

• Clear column headings.
• One observation per row.
• Consistent formats.
• Defined units of measurement.
• A clear reporting period.
• No unnecessary merged cells.
• Documented indicator definitions.

Facility identifiers help distinguish facilities with similar names. In real applications, use the authoritative identifiers and names supplied by the relevant organisation.

5.2.3 Importing an Excel file

1. Open Power BI Desktop.
2. Choose the option to get data.
3. Select Excel workbook.
4. Locate the workbook.
5. Select the relevant worksheet or table.
6. Preview the data.
7. Load the data, or transform it before loading.

The exact interface can vary by version. Power BI Desktop requires a compatible Windows computer. If you are learning on a phone only, study the concepts and use the Excel alternative at the end of this module.

5.2.4 Understanding data types

If a patient count is imported as text, calculations may not work as expected. Dates should be recognised as dates rather than unrelated text.

5.2.5 Cleaning data with Power Query

Power Query is the data preparation tool in Power BI Desktop. It can rename columns, change data types, filter rows, remove duplicates when appropriate, replace values, split columns, combine datasets and reshape data.

A tool cannot decide that every unusual value is wrong. If a facility reports 1,000 patients instead of its usual 100, investigate. Do not automatically replace 1,000 with 100.

5.2.6 Create a copy and document changes

1. Preserve the original source.
2. Record the cleaning steps.
3. Verify corrections where possible.
4. Document unresolved issues.
5. Confirm that the resulting dataset still represents the intended information.

Good dashboard design begins with good data preparation. Check the structure, formats, missing values, possible duplicates and indicator definitions before you build visuals.