It’s always fun to watch business consultants and IT companies search for the next buzzword - throwing candidates to the wall, hoping that one of them will stick and help them own the next big thing. (IBM is particularly adept at the game, rotating buzzwords every two years.)
Data scientists are waking up to the fact that combining multiple, relatively simple predictive models is often more efficient, when tackling a time series forecasting challenge, than using the latest super-duper neural network.
Datapred's machine learning software for direct material procurement helps companies save 3-5% off their commodity, energy and raw material costs year over year, through a unique combination of price predictions and constraint optimization.
We talk a lot with the data science teams of industrial companies, and it is striking how unaware they usually are of the gap between a good machine learning model and a production-ready machine learning application.
Seasonality - recurring but not necessarily periodic data patterns - is a staple of time series modeling. Since capturing true seasonality greatly enhances model accuracy, we wanted to share our thoughts and experience on the detection and modeling of such data patterns.
In a previous post, we explained the concept of cross-validation for time series, aka backtesting, and why proper backtests matter for time series modeling.
Our goal in this post is to discuss our standard strategy (beyond respecting basic time series modeling principles) for building accurate predictive models. We will use the example of commodity procurement optimization.
