Hi,
I hope you are enjoying the course so far. On Tuesday at 20:30 your time, we have the next applications talk.
Neil Headings is talking about building confidence in the outcome of modelling. This is a very interesting and important topic when you do big data and machine learning. With any algorithm, the computer does what you tell it to do and you get results based on what you put it. There are many famous example of people getting it horribly wrong. This is a big thing; especially since many people think that the computer is so smart that it must be right. For example, you might have heard about AI tools used to recognise shoplifters automatically when they enter the shop. In a famous example of getting it wrong, the network was trained on only men of a light skin colour and called everyone else a potential thief. There are many more of those. Think about prescriptions for mediation; if you test a new medicine on people that all weigh well over 120kg and are all 2m tall and above, you reach a conclusion on the amount of medication you need to give a patient. However, you will probably horribly overdose the 60kg woman who is only 1.50m tall. Or think of people being excluded from insurance because their risk profile looks too dangerous, but actually they are low risk. The examples are countless. Thinking about these things in advance is vital.
I hope you will join this talk. What you will learn is really key to be successful when applying data science in real life.
Please use this link to connect to the Teams meeting
Best wishes,
Jaap Velthuis.