This is a short deck I made recently for a 10-minute talk. It is a combination of Andrew Ng lectures with my own experience, and designed to be digestible in 10-15 minutes. There are actually quite a few other topics that I could also talk about (oh boy, I have a bag of tricks for practical Machine Learning: working with Big Data, special tricks for Deep learning, feature engineering techniques, etc…), but for the sake of consumability, I limited myself to those topics.
Bias-Variance tradeoff, although being a fundamental topic in ML, is actually not quite intuitive, and may take time to really understand. I believe it is one of the most crucial topic for successfully applying Machine Learning to real-world data.
Check it out and let me know what you think.