Great group discussion on Hidden Markov Models tag-teamed by Yaniv and Alisa. Walked through the coin flip example, creating the forward algorithm, and the posterior decoding. In this case it is assumed model and parameters are known: two coins, known ratios, known rate of swapping; goal is to guess which coin is being used in a given interval. Algorithm does quite well, as shown in Yaniv’s awesome graph.
As usual in the notebook, click on the graph to get link-through to the code.