{"id":9679,"date":"2022-06-24T08:47:52","date_gmt":"2022-06-24T12:47:52","guid":{"rendered":"https:\/\/demand-planning.com\/?p=9679"},"modified":"2022-06-24T10:02:07","modified_gmt":"2022-06-24T14:02:07","slug":"my-favorite-forecasting-model","status":"publish","type":"post","link":"https:\/\/demand-planning.com\/2022\/06\/24\/my-favorite-forecasting-model\/","title":{"rendered":"My Favorite Forecasting Model"},"content":{"rendered":"
\n

One of the questions I get asked most frequently is \u201cWhat is your favorite forecasting model?\u201d My answer is \u201cit depends\u201d because not all problems need a hammer. Sometimes you need a wrench or a screwdriver which is why I advocate having a forecasting toolbox that we can draw on to tackle whatever forecasting project arises.<\/strong><\/p>\n

When it comes to forecasting methods, we have everything from pure qualitative methods to pure quantitative methods, and everything in between. On the far left of the image below you\u2019ll see judgmental, opinion-based methods with knowledge as the inputs. On the far right, we have unsupervised machine learning \u2013 AI, Artificial Neural Networks etc. where the machine decides on the groupings and optimizes the parameters as they learn the test data. In between these two extremes we have na\u00efve models, causal\/relationship models, and time series models.<\/p>\n

All of the models should be in our toolbox as forecasters.<\/p>\n

But with dozens of methods available to you, how you decide which ones to use? There are cases when sophisticated machine learning will help you and there are cases when pure judgement will help but somewhere in between the extremes is where you\u2019ll find the models you\u2019ll need on a\u00a0 day-to-day basis.<\/p>\n

\"\"<\/p>\n

Picking The Right Model For A Particular Forecast<\/strong><\/h2>\n

The main thing is to have a toolbox full of different methods that you can draw on depending on the data available and the resources you have. We must balance 3 key elements when choosing a model:<\/p>\n

Time available<\/em><\/strong>: How much time do you have to generate a forecast? Some models take longer than others.<\/p>\n

Interpretability of outputs<\/em>:<\/strong> Do you need to explain how the model works to stakeholders? Outputs of some models are difficult to explain to non-forecasters.<\/p>\n

Data:<\/em><\/strong> Some models require more data than others and we don’t always have sufficient data.<\/p>\n

For example, putting together a sophisticated machine learning model and training it could take months to build, plus extra time for it to provide a useable output. When a forecast is needed now<\/em>, this kind of model won\u2019t help. Similarly you have little or no data as with new products, you may have to use judgmental methods.<\/p>\n

Balancing interpretability and accuracy is also key. There are models whose accuracy can be finetuned to a great degree but as we become more accurate, interpretability (explaining the rationale behind the number) often becomes more difficult. Artificial Neural Networks, for example, can be very accurate, but if you need to explain to partners in the S&OP process or to company execs how the model works and why they should trust it, well you might have some difficulty.<\/p>\n

Time series models like regression or exponential smoothing are much easier for stakeholders to understand. So what kind of accuracy do you need? Do you need 99% accuracy for a particular forecast, or is some margin of error acceptable? Remember that there are diminishing returns to finetuning a model for accuracy – more effort doesn’t necessarily provide more business value.<\/p>\n

This is why the best model depends on the context you\u2019re working in.<\/p>\n

 <\/p>\n