John Mount recently led a time series workshop for AI+ training hosted by ODSC, showcasing a graph depicting the growth of Python views on Stack Overflow. The late 2017 prediction was deemed reasonable due to good methodology and accurate future data matching the forecast. Mount also discussed using the radical Bass product diffusion model, which assumes all things end abruptly. By fitting this model with Bayesian inference through Stan, he discovered interesting trends and changing growth patterns. While not a perfect fit, the models effectively predict future growth until new data becomes available. The data and R code for these graphs can be found on Mount’s website.
https://win-vector.com/2025/03/14/changing-forecasts-for-python-on-stack-overflow/