Explore how shifting from traditional estimating techniques to probabilistic forecasting can revolutionise your approach, offering precision, efficiency and direct cost benefits.
Something to consider, from a Deming perspective: As useful as Monte Carlo sims are for creating probabilistic models based on past data, they have a critical shortcoming in that they don't tell us if the system creating the data is itself stable and predictable. Forecasting against an unstable system will contribute to wide probabilistic forecasts that wouldn't improve because the special causes of variation that are contributing to the instability aren't even known about or understood. A different statistical tool is required that can help us find signals of instability in noisy data so we find out why and what, if anything, we can do about them. Enter: the Process Behaviour Chart.
Something to consider, from a Deming perspective: As useful as Monte Carlo sims are for creating probabilistic models based on past data, they have a critical shortcoming in that they don't tell us if the system creating the data is itself stable and predictable. Forecasting against an unstable system will contribute to wide probabilistic forecasts that wouldn't improve because the special causes of variation that are contributing to the instability aren't even known about or understood. A different statistical tool is required that can help us find signals of instability in noisy data so we find out why and what, if anything, we can do about them. Enter: the Process Behaviour Chart.
I explain my thinking on this with respect to forecasting software development throughput in my Red Bead Redux post, here: https://digestibledeming.substack.com/i/120737445/the-punch-line-dont-forecast-an-unstable-system