In the past month or so we've been giving talks and workshops on one of our projects to predict customer spend on different digital marketing services.  Briefly, the project aimed to use what's called a "probabilistic model" to output a realistic spend range for a given customer in a certain region, certain industry and for a particular managed services product.  I won't go into too much detail here as there's a lot of information available in the talk materials.

 One in-depth talk was given at the Open Data Science Conference (ODSC) in London and a complete walkthrough was presented at the PyData conference in NYC.

 You can see the slides for the ODSC conference here .  The walkthrough is presented in a Jupyter Notebook here.  If you want to play with notebook yourself, you'll either need to use the requirements.txt file in the github repository or download the Docker image according to the instructions on the main page.

 I'm planning on putting together a short series walking through, at a high and slightly less-technical level, how these models work and how they're being applied at ThriveHive.  Stay tuned!