In this session, I will live code a tool that detects outliers (or fake news) in time series. My goal during this session is to demystify the maths behind these analyses, focusing on the context and explaining the difference between recommended methodologies. I will focus on the questions that each tool helps us answer, rather than giving an ""one-size-fits-all"" solution. I will also talk about data quality assessment, clustering techniques and prediction methods. In the end, we will have a jupyter notebook which will have a selection of tools and visualisation charts, so that people can take this away and expand and critically assess the solutions.
I will make a github repo where I will put all the information required (datasets, required installations etc). This will be available no later than the 27th of December.
The audience does not have to code along with me if they don't want to, they can watch/ask questions/shout ""you made a typo theeereee"", in the end I will share the notebook anyway :)

I can promise laughter, mild excitement, and a lot of facepalms.
This session is beginner-friendly.