Digital, AI and Cyber Security
Data and Analytics
Advanced Visualisation with Power BI Training
Building a report somebody acts on, where the model is right and the chart does not mislead.
- Duration
- 5 days
- Contact hours
- 30
- Level
- Advanced
- Fee from
- On enquiry
Overview
Most Power BI problems look like visual problems and are model problems. A measure that double counts, a relationship pointing the wrong way, a date table that is really a column, and filter context nobody has thought about will produce a chart that is confidently wrong. The visual layer then has its own failures: dual axes that imply a relationship, truncated scales, and a page holding thirty numbers and no answer. This week covers both, in that order, on the delegate's own data where they can bring it.
Who should attend
Analysts and reporting staff building Power BI reports, finance and operations staff who maintain dashboards, and managers who commission them.
What you will be able to do
- Build a model with a proper date table and relationships that point the right way.
- Write measures that survive being sliced by something unexpected.
- Choose a chart that answers the question rather than decorating the number.
- Design a page around one decision rather than around available fields.
- Diagnose a figure that is wrong without rebuilding the report.
How it is taught
Short input, then work, across 5 days of roughly 6 teaching hours each. Exercises run on situations delegates bring rather than on a case study written for a different organisation. Cohorts are capped so that everybody is asked something.
Available as classroom, in house.
Programme outline
Day 1
The model decides the answer
- Star schema, and why a flat table stops working
- Relationships, direction and ambiguity
- A real date table, and what breaks without one
- Granularity, and measures that double count
Day 2
Measures
- Filter context, and the slice that changes the answer
- Time intelligence that is right at period boundaries
- Ratios, averages and the denominator problem
- Testing a measure against a known total
Day 3
Choosing the visual
- The question first, the chart second
- Comparison, composition, distribution and trend
- Dual axes, truncated scales and other honest mistakes
- Tables, and when a table is the right answer
Day 4
Designing a page
- One decision per page
- Hierarchy, so the headline is read first
- Interaction, drill through and what a reader will actually click
- Colour used to carry meaning rather than for decoration
Day 5
Running it in production
- Refresh, gateways and the failures that go unnoticed
- Row level security, and testing it as a user
- Performance, and what makes a report slow
- Documenting a report so somebody else can own it
What you leave with
- A reviewed and corrected model for one of the delegate's own reports.
- A measure set with the filter context behaviour documented.
- A redesigned page built around a single decision.
Dates and fees
| Dates | Location | Format | Fee |
|---|---|---|---|
| 23 to 27 November 2026 | Dubai, United Arab Emirates | Classroom | US$1,500 |
| 30 November 2026 to 4 December 2026 | Cape Town, South Africa | Classroom | R15,000 |
| 7 to 11 December 2026 | Kampala, Uganda | Classroom | US$1,500 |
| 14 to 18 December 2026 | Lagos, Nigeria | Classroom | US$1,500 |
Fees are per delegate and cover tuition, all programme materials and the certificate. Travel, accommodation and visas are arranged by the delegate. Registration closes two weeks before a programme starts.
Certificate
A GATI certificate recording what was attended, issued after the programme has run and attendance has been taken. Attending at least 80 per cent of the days earns a certificate of completion; attending less earns a certificate of attendance stating exactly what was attended. Every certificate carries a code that anybody can check against GATI's own record.
It records participation. and GATI does not claim one it does not hold.