Statistical forecasting is one of the main pillars of ACT Operations Research approach
Every business prediction starts with statistical forecasting!
When it comes to minimize forecasting errors and keeping risks at bay, relying on the right techniques is what makes the difference. Large data sets and multiple variables can lead to overfitting. Good forecasts empower to manage volatility risks rather than being a victim.
FAQ - Frequently Asked Questions
- How good are my forecasts?
- How do you assess the forecast accuracy?
- How are seasonalities and trends managed?
- How do low rotation and discounted products come through your models?
- Are data errors automatically spotted and managed?
- How do you manage cannibalization and gripping circumstances?
- Can I set a sustainable risk range?
- Are weather and other influencing variables considered?
- How are stock-outs considered?
- How are special sales considered?
- Can vertical drilling (top-down / bottom-up) be performed?
- How do you perform forecasting on short-life products?
Techniques:
At ACT Operations Research we use popular and advanced techniques.
Some of the former are the following:
- autoregressive, moving average, exponential smoothing (e.g., (double, triple), Box-Jenkins, Holt-Winters
- ARMA, ARIMA
- Support Vector Machines (SVM) and Support Vector Regression (SVR)
- Neural Network
Third-parts technologies:
|
