Which Forecasting Model/ Forecasting Stratergy is suitable for Steel Plant

[former_member224276]

Hello Friends,

Could you pl suggest me Which Forecasting Model/ Forecasting Stratergy is suitable for Steel Plant.Any one having the steel plant experiance pl suggest me .

Edited by: sivaram2411 on Jan 19, 2011 6:37 AM

[Former Member]

Also read about Ex-Post Forecast http://help.sap.com/saphelp_scm50/helpdata/en/ac/216b77337b11d398290000e8a49608/content.htm

[Former Member]

Hi Sivaram,

I haven't worked in Steel but worked in Cement industry. Cement is another input to construction industry, just like steel. In India, both have seasonal demand. But I would give a generic suggestion here, though you can check with your network in other steel companies having similar product offering as yours, about the forecasting model they follow.

Choice of forecasting model is not industry dependent. Because within the same industry, each player has a different product mix, different set of customers and in different geograhies. Forecasting model is selected based on the analysis of historical data. So you can take last 2 years available data to start with.

Demand Planning provides the following statistical forecasting models: constant model, trend model, seasonal model, trend and seasonal model, Croston method with exponential smoothing, linear regression, and causal models with multiple linear regression (MLR).

Time series models are the most commonly used forecasting methods.

The flow chart is like this:

Raw Data collection --> Data analysis --> Analyzed data --> Forecasting Model selection

You can use plotting as a method of Data analysis,

We need to find the best forecasting model, so that it gives us the most accurate demand. There is no systematic approach for the selection of a forecasting model and you need to find the best for you by trial and error.

The following factors impact the model selection: (1) Forecast Horizon (2) Data Availability / Requirements (3) Pattern of Past Data (4) Level of Detail in the Forecast (5) Cost Factors (6) Accuracy and (7) Ease of Application.

So use the historical data, analyse it, try using the available forecasting methods one by one to forecast for a period during which the actual data is already available with you and find the one which is best matching the actual figures. The model giving most accurate results, or the one with least forecast error, would be the one you are looking for.

[Former Member]

Hi Sivaram,

The example history you have given is not constant or seasonal or linear.

To determine the best model in interactive planning for a small amount of representative data, you have 2 options :

1) Manually, by forecasting using various models and recording the errors produced by each model. You can select the model that results in the smallest errors. For more information about forecast comparision see

http://help.sap.com/saphelp_scm70/helpdata/en/a6/fbe54777d111d3984a0000e8a49608/frameset.htm

2) Automatically, by selecting one of the two automatic model selection procedures.

Thanks and Regards,

Nithin.

[former_member224276]

Hi Friends,

Those models are not suitable for this type of history.

For example i will provide 12 months history

2022 1391 1768 3218 3123 942 1820 2241 563 687 870 1529

any one having experience in steel plant pl suggest me which model is suitable.

more comments pl.

[Former Member]

Hi,

From my point of view, I dont think we need to use any special forecast model for this. Average/Mean would do the job.

Regards,

Guru Charan.

[Former Member]

Hi Siva

I will suggest you to go from following link.

http://www.metal2011.com/data/metal2010/sbornik/lists/papers/128.pdf

Please let us know , if this helps you.

Thanks

Amol