Ventana software support forum

My model's objective is to set safety stock based on the prediction of the next transportation time. Transportation time is influenced by multiple factors and exhibits seasonal patterns; it is longer in the summer and winter seasons and shorter in the spring and autumn seasons. Therefore, I hope to embed a seasonal forecasting module (using the Holt-Winters method) into the model. This module will use historical transportation time data to predict the next transportation time.

Additionally, in the variables I have set for statistical analysis of transportation time, there are instances with a value of 0, indicating no transportation tasks. In such cases, how should I select non-zero data for forecasting?

I have attempted first-order exponential smoothing, but I am uncertain if this approach is reasonable. My basic model is attached for reference.

Looking forward to your insights.

Best, Xuanao

Start by adding units to your model. I would not bother simulating or doing anything else until your model passes this basic quality check. I can immediately see a problem (you've got a variable called month).

I think the month variable is intentional, for creating a seasonal input.

Since the seasonal behavior has 3 episodes, switched in an IF THEN ELSE statement, I think you could create the same thing for seasonal smoothing - have 3 smoothing structures, one active during each episode.