This is the final post of the mini series on Eurovision analysis. If you haven’t read part one or part two I would recommend you read them first.

Orchestration

Before we cover the reporting and analysis, I thought it worth briefly covering how all the notebooks are orchestrated. To control notebook orchestration there are a few different options within Fabric, runMultiple with a DAG, Airflow, or a data pipeline.

There are a few pros and cons for each option, such as cost, easy of setup, etc. For this simple project I went for a data pipeline as it was the easiest to quickly setup. As for the pipeline itself it consists of a few notebook activities linked with dependencies.

This is part of a mini series on analysing Eurovision data with Microsoft Fabric. If you missed part one you can read it here. In part 1 I briefly covered the initial data load via the Youtube API and web scraping. So now we need to enhance this data to be ready for final analysis.

Mapping videos to countries

In the initial load of the videos table, the title for the video was returned by the API. However, in order to do meaningful analysis later we need to extract the country of the video. After a few attempts at working out a dynamic way to do this, such as going of flag emojis. However, due to the wide range of variations of title I eventually settled on just having a list of regex with one value for each country that has competed in Eurovision.

With Eurovision 2025 only a few weeks away I thought it was time for me to finish some experimenting I had started last year around looking into bringing Eurovision data into Fabric. See what insights we can gain, and maybe if we can predict the winner!

The Plan

As with any event like Eurovision there is a wide range of possible data sources, in order to get a mix of data I decided to start with comparing data from the official music videos on Youtube with the results of the previous contests. This included both simple statistics such as views and likes on each video but also the sentiment on a sample of the comments using some of the AI functions built into Fabric.

In a recent project I had to add field parameters to a Direct Lake model, as currently field parameters can’t be created in the Power BI Service at the moment you have to either use a tool like tabular editor or a recent preview feature in Power BI desktop. However, I found the documentation for this preview feature very patchy so I hope I can provide a clear step by step guide.

Quick background first, what are a non-breaking spaces and why would you want to use them?

A non-breaking space is a character which looks the same as a normal space, however it prevents software that is rendering the text from inserting a linebreak where that space is. This is useful if you want to control exactly where text may be moved onto a new line, especially with different sized screens etc. The key point for Power BI and Power Query is that these non-breaking spaces have a different unicode value to the normal space.