April 9, 2019

Machine Learning Advice needed


Hello IH

Building out our MVP, hoping to be live next week in our first POC.

Our app is going to ingest and store loads of transactional data, ie point of sale transactions.

Where would you look for an AI Machine learning set up that can ingest and begin surfacing insights over time.

I realize it would need training and a data scientist, but if I get it in the right place to start, that can help.

Much appreciated

  1. 1

    I'd recommend getting more specific on what kind of insights you're looking to surface.

    I have a lot of experience doing machine learning consulting and searching for these kinds of "insights" (including with transactional retail data) and honestly think you can use libraries and copy / paste from stackoverflow for most of the "AI Machine learning" part provided you know what kind of insights you'd like to get.

    The most obvious ones are:

    • product recommendation (search for "collaborative filtering") - basically this just says "products who bought X also bought Y"
    • forecasting sales (take a look at this: https://github.com/fastai/fastai/blob/master/courses/dl1/lesson3-rossman.ipynb)
    • logistics / supply chain optimization (predicting when items will sell out so you don't run out of stock)
    • fraud detection (might not be relevant but essentially this is just anomaly detection)
  2. 1

    Hey Smithy, I recently built https://mlrequest.com for this problem. You can create production models with little effort. All models use online learning, so whenever you feed new data to a model it will learn incrementally over time.