PlayML notebook 'Pricing under Uncertainty' now online

At Datasparq we publish interactive explorations of machine learning at our PlayML site. I’ve written a new one!

Pricing under Uncertainty

Whenever quantitative techniques are used to describe the real world, a toy model is a deliberately simplistic representation of that world. Toy models can be used to concisely explain a mechanism without introducing too many complicating factors. This is an analogy of how toys are used the real world: a model kitchen, for example, is a deliberately simplified version of the real thing, enabling children to play at being grown-ups without setting the house on fire.

This is what PlayML is all about: playing with toy models of AI and ML techniques, without setting the (metaphorical) house on fire.

In this PlayML notebook, we’re going to play the part of a business that makes model kitchens for kids, trying to set a price for our product. As we’re in the model business already, we will build a toy model of how the technique of Thompson sampling can be used to make this pricing decision.

See more here!