Lean Kanban North America 2016

After learning and understanding the benefits of using a probabilistic approach to forecast when a feature (or a set of them) might be completed, Monte Carlo simulations were one of the first things I worked on adding to my analysis tool belt. Conceptually, I find it to be a simple technique, and relative low cost effort in comparison to the potentially powerful insights it provides beyond single point or non-probabilistic based range estimates.

However, as with many analysis methods and in particular one where a model is used, there are assumptions made quite often implicitly, if not explicitly, or possibly even unknowingly. In this workshop we’ll see examples of just how important it is to be aware of these assumptions in your context.

I’ll start with a quick overview of a simple Monte Carlo simulation I developed using “real-world” data and identify some of my initial assumptions. Next, we’ll see how the forecast compared to actual results as part of my efforts to validate the model design and assumptions. And we’ll conclude with how this validation effort resulted in new learning and understanding of how important those assumptions can be in the value and benefit you obtain from a forecast of this nature.

While no formal previous statistics knowledge is required, as a “heads-up” in addition to Monte Carlo simulations our discussion will assume some basic familiarity with these concepts: random numbers, scatter plots, percentiles, and histograms.