SCM 301 Chap 9

Seasonal Adjustments Four-Step Procedure:

1. For each of the demand values in the time series, calculate the corresponding forecast using the unadjusted forecast model. 2. For each demand value, calculate (Demand/Forecast). If the ratio is less than 1, then the forecast model over forecasted; if it is greater than 1, then the model under forecasted. 3. If the time series covers multiple years, take the average (Demand/Forecast) for corresponding months or quarters to derive the seasonal index. Otherwise, use (Demand/Forecast) calculated in Step 2 as the seasonal index. 4. Multiply the unadjusted forecast by the seasonal index to get the seasonally adjusted forecast value.

Forecasting is used to determine:

1. Long-term capacity needs 2. Yearly business plans 3. Shorter-term operations and supply chain activities

Qualitative Forecasting Methods

1. Market surveys 1. Structured questionnaires submitted to potential customers 2. Panel consensus forecasting 1. Experts come together to develop forecasts 3. Delphi method 1. Experts work individually to develop forecasts 4. Life-cycle analogy method 1. Identify the time frames and demand levels of different stages of new product or service 5. Build-up forecasts - Experts familiar with specific market segments estimate the demand within these segments

A set of business processes, backed up by information technology, in which supply chain partners agree to mutual business objectives and measures, develop joint sales and operational plans, and collaborate to generate and update sales forecasts and replenishment plans.

Collaborative Planning, Forecasting, and Replenishment (CPFR)

______ are used to develop, evaluate and change forecasting models as needed. - With enough demand history, a package could quickly evaluate alternative forecasting methods for each item and select the model that best fits the past data. - Some packages can use MFE, MAD, and MAPE or tracking signal criteria to flag a poor forecasting model and automatically kick off a search for a better one. Others can develop multiples forecasts for a single item.

Computer-based forecasting packages

laws of forecasting

Law 1: Forecasts Are Almost Always Wrong (But They Are Still Useful). ■ Law 2: Forecasts for the Near Term Tend To Be More Accurate. ■ Law 3: Forecasts for Groups of Products or Services Tend to Be More Accurate. ■ Law 4: Forecasts Are No Substitute For Calculated Values.

______ are used to assess how well an individual model is performing or to compare multiple forecast models to one another.

Measures of Forecast Accuracy

Demand Patterns

Randomness Trend Seasonality

A quantitative forecasting model that uses a time series to develop forecasts

Time series forecasting models

An expanded version of the exponential smoothing model that includes a trend adjustment factor.

adjusted exponential smoothing model

The difference between Delphi method and Panel Consensus forecasting is_____________ a. Delphi uses time frames b. Only Panel Consensus uses experts c. Delphi are individuals vs. Panel which is a team d. Only Delphi uses experts

c. Delphi are individuals vs. Panel which is a team

A class of quantitative forecasting models in which the forecast is modeled as a function of something other than time.

causal forcecasting models

- Overall market demand - Firm-Level demand

demand

Which of the following, is not part of the Laws of Forecasting? a. Forecasts are almost always wrong b. Forecasts for groups of products or services tend to be more accurate c. Forecasts are no substitute For calculated values d. Forecasts for the near term tend to be more accurate e. Forecasts are usually have very limited usefulness because they are inaccurate

e. Forecasts are usually have very limited usefulness because they are inaccurate

This forecast metric is the only one that does not target a value of zero. a. MAPE b. MAD c. Tracking signal d. MFE e. None of the these

e. None of the these

A special form of the moving average model in which the forecast for the next period is calculated as the weighted average of the current period's actual value and forecast.

exponential smoothing model

True or False Quantitative modeling is the most opinion based approach to forecasting.

false

True or False Your forecast in control if your tracking signal is -5.

false

An estimate of the future level of some variable

forecast

The less randomness in the time series data, the _____ the α value should be

higher

The simplest time series model which uses demand for the current period as a forecast for the next period

last period model

- A statistical technique that expresses a forecast variable as a linear function of some independent variable. - Can be used to develop time series and causal forecasting models.

linear regression

The greater the randomness in the time series data, the____ the α value should be.

lower

- A time series forecasting model that derives a forecast by taking an average of recent demand values.

moving average model

A generalized form of linear regression that allows for more than one independent variable

multiple regression

- Forecast prices for key materials and services

price

Forecasting techniques based on intuition or informed opinion. - Used when data are scarce, not available, or irrelevant. - Opinion driven modeling

qualitative forecasting techniques

Forecasting models that use measurable or historical data to generate forecasts. - Time Series and Causal models - Data driven modeling

quantitative forecasting models

Unpredictable movement from one time period to the next

randomness

A repeated pattern of spikes or drops in a time series associated with certain times of the year.

seasonality

Repeated patterns or drops in a time series associated with certain times of the year.

seasonality

- Number of current producers and suppliers - Projected aggregate supply levels - Technological and political trends that might affect supply

supply

A series of observations arranged in chronological order

time series

Long-term movement up or down in a time series

trend

True or False In Exponential Smoothing, the greater the randomness in the time series data, the lower the α value should be.

true

True or False You have good historical data so you will likely want to use a quantitative model to forecast

true

A form of the moving average model that allows the actual weights applied to past observations to differ.

weighted moving average model