Forecasting-Process Forecasts are estimates of timing and magnitude of the occurrence of future events Key functions: An estimation tool A way of addressing the complex and uncertain environment surrounding business decision making. A tool for predicting events related to operations, planning and control. A vital perquisite for the planning process in organizations.
Why do we forecast… Dynamic and complex environment Short term fluctuations in production Better materials management Rationalize man-power decisions Basis for planning and scheduling Strategic decisions
Focus-Forecasting-Introduction Bernie Smith- Servistardivision of TruValue Two Principle theory All complex forecasting models are not always better than simpler ones. No single technique for products and services Simple techniques that work on past data also helps in developing forecasts about future as well Reasonable approach for short term (period less than a year)
Some examples of Forecasting Rules are as follows:
Whatever is sold in last 6 months is what will be the sales in
Current year’s quarter wise sales can be predicted by last
year’s quarter wise pattern
Next three months sales will be 10 percent more than last three months sales
Percentage change in sale levels will be same for all the quarters this year
For non-seasonal items and spare parts with weak and irregular demand
The sales for the item in the next quarter will be the same as the actual sales for the last quarter F = Forecast for the item over the next quarter Q1 = Actual Sales over the most recent 3 months° (example: 100) °the first quarter in the past counting backwards from now F = Q1 ; F = 100 for the next quarter This formula looks modest but is nonetheless surprisingly robust for all types of non-seasonal items, whether their demand be strong or weak
Contd.. For items with an irregular demand history
Sales for the item in the next quarter will be half of actual sales over the last 6 months F = Forecast for the item over the next quarter Q1 = Actual Sales over the most recent 3 months° (example : 100) °the first quarter in the past counting backwards from now Q2 = Actual Sales over the 3 months before that ° (example : 150 ) ° the second quarter in the past counting backwards from now ( Q1 + Q2/2)=(100 + 150)/2 = 125 for the next quarter This formula generates a reasonable forecast despite a demand history needing to be corrected
Sales for the item in the next quarter will follow the percentage increase or decrease with respect to last year. F = Forecast for the item for the next quarter. Q4 = Actual Sales of the next 3 months for last year° (example: 100) ° The fourth quarter in the past counting backwards from today Q1 = Actual Sales over the most recent 3 months° (example : 100) °the first quarter in the past counting backwards from now Q5 = Actual Sales of the corresponding quarter last year° (example : 150) ° The fifth quarter in the pas counting backwards from now
F = Q4 x Q1/Q5; 100 x 100/150 =100*.67= 67 for the next quarter
This formula is useful for following the market for many items, as long as their seasonality is moderate and the corresponding quarter of last year was not zero
Advantages " A Complex answer is just used to hide the fact that the answer has not yet been found - complex answers don’t work". B. Smith
Simple approach and easy to understand
Helps people in forecasting seasonality, trends, items with sporadic history and other demand conditions
Selects the best option which results in the least error out of varied forecasting models
It lays emphasis on simulation rather than the optimization
It can even include methods like exponential smoothing, if desired
It can regenerate forecast values without impairing the performance of hardware.
Small sample size. Only three monthly forecasts.
Ad hoc system with no theoretical basis to aid analysis or understanding.
Impossible to compute confidence intervals, regions of stability for the forecasts or
other standard analytical tools
No way to predict how Focus Forecasting should perform compared to any other forecasting system.
Examples of companies using focus forecasting Oracle E-Business Suite Manufacturing and Supply Chain Management Currently offers two products- Master Scheduling/MRP: offers single-organization, unconstrained planning of material and resources Advanced Supply Chain Planning: offers multi-organization planning and the option of constraint-based and optimized plans.
Contd… process involves the recognition of demand netting of those requirements against available and scheduled quantities generation of recommendations to meet those requirements proceeds top-down through the bill of material
Contd.. Considers alternate forecast scenarios, and so the forecasts are stated in a Master Demand Schedule (MDS) determine how much demand satisfied from existing stock or existing orders Oracle Demand Planning generates forecast data. Oracle Inventory and Master Scheduling/MRP provide basic methods to generate forecasts from historical data one forecast typically contains multiple items; each item has multiple entries. For ease of use and to control forecast consumption, forecasts are grouped into forecast sets and forecasts and forecasts sets are identified by unique names
Generating Forecasts from Historical Information Statistical Forecasts: can span multiple periods and can recognize trend and seasonality. Focus Forecasts: examines five different forecast models against past history, determines the model that would best have predicted the history, and uses that model to generate a forecast for the current period
Generating Forecasts defining a forecast rule forecast method (statistical or focus), and the sources of demand in the Generate Forecast window Name of the forecast you want to populate. Forecast rule Selection criteria to identify the items An overwrite option Start date and cutoff date Post this Oracle uses Open Forecast interface and forecast entries API to integrate with other systems
Other companies WorldPac Carquest Auto parts Digi-Key guestsupply.com usballoon.com