How Does Hidden Markov Models explained in plain English - Outspoken Market Work?

A Hidden Markov Model is like a detective game where you guess what someone is doing based on clues they give you.

Imagine your friend is playing hide-and-seek, but you can’t see them, only hear them. They might say “I’m near the tree!” or “I’m behind the couch!” These are their clues. But you don’t know if they’re really near the tree or just pretending. That’s like the hidden part of a Hidden Markov Model, what’s actually happening, but not directly visible.

Now imagine your friend has a routine: sometimes they hide near the tree, and other times behind the couch. This is their state, and it changes over time. You might hear them say “I’m near the tree!” three times in a row, that could mean they’re still there or just moving around. That’s like how Hidden Markov Models use probabilities to guess what state your friend is in.

Every clue (like “near the tree”) gives you a hint about their current state, but not all clues are perfect, sometimes they might say “I’m near the tree!” even if they’re actually behind the couch. The model uses these clues and probabilities to make its best guess about what your friend is really doing.

It’s like being a detective who never sees the person but listens closely and keeps track of their patterns, that’s how Hidden Markov Models work!