Statistical or univariate forecasting predicts future demand based on historical data. Unlike causal forecasting, it does not take other factors into account.
Univariate estimation provides methods that recognize fundamental time series patterns as the basis for forecasting:
Constant: Demand changes very little from a fixed average value.
Trend: Demand consistently falls or rises over a long period, with only occasional deviations.
Seasonal: Demand shows periodically recurring peaks that differ significantly from a stable average value.
Seasonal trend: demand shows periodically recurring peaks, but with a consistent increase or decrease in the average value.
Intermittent: Demand occurs only during certain periods.
Causal Estimation
Multiple linear regression (MLR) allows you to incorporate causal variables (such as climate conditions, price, and advertising) into the forecasting process. MLR investigates the historical impact of these variables on demand to generate a forecast. You can set up different scenarios for the causal variables to simulate possible developments and thus identify potential risks and opportunities.
Compound Estimate
The primary goal is to leverage the strengths of each method to create a single "one-number" estimate. By combining the estimates, the business analyst aims to develop the best possible forecast. Composite estimates from various methods have proven to outperform the individual estimates of any of the methods used to produce the composite. However, the planner can also allow the system to automatically select the individual estimate that yields the lowest statistical error.
Life Cycle Planning
A product's life cycle consists of different stages: launch, growth, maturity, and end.
You can represent the launch, growth, and termination phases using phase-in and phase-out profiles. The phase-in profile mimics the rising sales curve you would expect the product to show during its launch and growth phases, while the phase-out profile mimics the falling sales curve you would expect the product to show during its discontinuation phase. For new products, using historical data of the predicted products (like previous ones) as a forecasting basis has proven proven. This can be done using what is known as similar modeling. Using the central maintenance example for interchangeability relationships, the above profiles can be automatically generated and assigned. Lifecycle planning is considered in statistical forecasting and is studied at both detailed and aggregate levels.
Promotion Planning
In Demand Planning, you can schedule promotions or other special events individually.
You can use promotional planning to record one-off events like millennium celebrations or recurring events like quarterly advertising campaigns. Other examples of promotions include trade shows, trade discounts, dealer allowances, product displays, coupons, contests, freestanding inserts, and non-sales-related events such as competitor activity, market intelligence, up/down economic trends, strikes, and hurricanes.
Data Processing
Demand planning should include all available information regarding past sales, budgets, strategic company plans, or sales targets. This data can come from various sources and can be transferred from any source.
Collaborative Demand Planning
All Demand Planning data is available online to involve internal or external partners in the planning process. This ensures all partners agree on defined quantities, horizons, and conditions.
Planning with Bills of Materials
In addition to planning demand for a product, you can also estimate dependent demand at different planning levels by inflating bills of materials. This is important, for example, when you need to plan demand for a group of products sold in a promotional activity.
For example, this can be used for a set of several finished products (which may also be sold separately). The planning demand for the set creates dependent demand that can be combined with independent demand for the individual products. The overall demand for the product can then be used for sourcing, production, and supply planning.
Characteristic-Based Prediction
In Demand Planning, you can create a forecast based on the specifications of customizable end products. For example, in the case of a car, you can plan the color, engine, and climate control features. You can also forecast the demand for a combination of various features, thus accounting for the interdependence of demand for those features.
Customer Forecasting
Customer Forecast Management is used to receive and analyze incoming customer forecast data and make necessary adjustments before releasing it to Demand Planning for further planning. Analyzing forecasts allows the vendor to detect trends and changes in customer demand and incorporate this information into replenishment planning. Customer Forecast Management enables a higher response to fluctuations in demand and also contributes to preventing inventory overruns.