Forecasting Methods Quiz

Introduction to Forecasting

• Managers constantly strive to reduce uncertainty and enhance future estimations.
• Forecasting serves as the primary tool for achieving this objective.
• Subjective methods, like intuition and experience, are sometimes used by firms.
• Several quantitative techniques, including moving averages, exponential smoothing, trend projections, and least squares regression analysis, are employed by businesses.
• Forecasting involves methodical steps:
  • Defining the forecast's objective
  • Selecting the items to be forecasted
  • Determining the forecast's time horizon
  • Choosing the appropriate forecasting model or multiple models
  • Gathering necessary data
  • Validating the forecasting model
  • Making the forecast
  • Implementing the forecast results

What is Forecasting?

• Forecasting is a process of estimating or predicting future demand through past and present events.
• It provides insights into potential future events and their impact on businesses.
• Forecasting is considered a blend of art and science in estimating future events (Heizer and Render, 2010).
• It involves a systematic approach to inferring the future based on known facts (Louis Allen).

Strategic Importance of Forecasting

• Supply Chain Management involves all activities that ensure the right product, at the right price, and in the right place.
• Apple Inc. exemplifies effective supply chain management by controlling diverse aspects of its global system.
• Human resources (HR) demand forecasting predicts future human resource needs to ensure the right number and quality of personnel. HR functions, such as hiring, training, promotions, and lay-offs, rely on this forecast.
• Capacity from an organizational perspective encompasses all elements—physical resources, ideas, and people—required to achieve the organization's mission and meet demand.

Benefits of Forecasting

• Aids in inventory and material management.
• Improves employee relations.
• Optimizes the utilization of available resources.
• Enhances customer satisfaction.

Forecasting Models

• Models are classified into qualitative and quantitative categories.
• Qualitative methods include Delphi Methods, Jury of Executive Opinion, Sales Force Composite, and Consumer Market Surveys.
• Quantitative methods are further categorized into time-series and causal techniques.
• Time-series methods, such as moving averages and exponential smoothing, identify trends and predict future values based on historical data.
• Causal models employ variables or factors potentially affecting a forecasted quantity. Regression analysis serves as the standard quantitative approach for causal modeling.

Time-Series Models

• Time-series models anticipate future values based exclusively on historical patterns.
• Methods including moving averages, exponential smoothing, trend projections, and decomposition utilize past data to make predictions.
• Regression analysis underpins trend projections and some decomposition models.

Causal Models

• Causal models identify and model the relationships between factors influencing the forecasted quantity.
• The goal is to establish a statistically sound relationship between the variable being forecast and relevant influencing factors.
• Regression analysis is the most prevalent technique employed in causal modeling.

Qualitative Models

• Qualitative models incorporate judgment or subjective factors, proving valuable when accurate quantitative data acquisition is challenging.
• Qualitative methods like Delphi, Jury of Executive Opinion, and Sales Force Composite incorporate expert judgment.
• Consumer market surveys gather input from potential or current customers about purchasing plans.

Scatter Diagrams

• Scatter diagrams visually represent the relationship between variables in a time series, aiding forecast construction.
• Wacker Distributors utilized scatter data to forecast sales for various products.
• By examining plotted data points, patterns can be identified; if the relationship between variables appears linear, a trend projection might be employed to predict future outcomes.

Measures of Forecast Accuracy

• Forecasts are evaluated against actual values.
• Common error measures include Mean Absolute Deviation (MAD), Mean Squared Error (MSE), and Mean Absolute Percent Error (MAPE).
• Bias, representing the average error, indicates whether forecasts tend towards high or low values.
• Naïve forecasting models serve as baseline comparisons for evaluating the accuracy of more sophisticated methods.

Hospital Days Forecast Example

• Ms. Smith, forecasting total hospital inpatient days, reevaluates her model using provided actual data.
• MAD, MSE, and MAPE are used for model evaluation.

Time-Series Forecasting Models

• A time series compiles evenly-spaced events.
• Time-series forecasts solely utilize past data values.

Moving Averages

• Moving averages produce forecasts when demand remains relatively stable over time.
• The forecast is the average of recent data values from the time series, replacing the oldest data point as new values become available.

Weighted Moving Averages

• Weighted moving averages assign varying weights to past data points, emphasizing recent data.
• The technique is particularly beneficial when a trend or other pattern is emerging in the data.

Exponential Smoothing

• Exponential smoothing is a data-smoothing technique that places greater weight on the most recent data points.
• The forecast for a period consists of the previous forecast adjusted for the error.
• A smoothing constant (0 ≤ α ≤ 1) controls the relative importance given to recent versus prior data in the forecast.

Selecting the Smoothing Constant

• Choosing the proper smoothing constant (α) is critical in creating a reliable forecast.
• Experimentation to find the smoothing constant (α) with the lowest mean absolute deviation (MAD) is commonly employed.

Trend Projections

• Linear regression generates a trend line from observed data.
• Future forecasts are calculated by projecting this line into the future.
• The least squares method determines the trend line that minimizes errors between predicted and actual values.

Midwestern Manufacturing Company Example

• The data shows a trend in electrical generator sales.
• Using simple linear regression, forecasts for the next two quarters can readily be made.
• Data analysis helps to decide on the best model and its forecasting accuracy.

Decomposition of a Time-Series

• Time series consist of components including trend, seasonality, cycles, and random variations.
• Multiplicative models express demand as the product of trend, seasonal, cyclical, and random factors.
• Additive models represent demand as the sum of these components.

Seasonal variations

• Recurring fluctuations suggest the need for seasonal adjustments within trend lines.
• A seasonal index quantifies how a particular season compares with the average.
• This index calculates the ratio of the average value for a certain season to the average across all data points when no trend is present.

Regression with Trend and Seasonal Components

• Multiple regression methods forecast a trend or seasonal pattern by using time period and dummy variables to capture seasonal effects.
• This approach builds a model for forecasting.

Monitoring and Controlling Forecasts

• Tracking signals monitor forecast accuracy.
• A tracking signal exceeding predefined limits indicates a need for analysis or adjustments to the forecasting model or procedure.
• A standardized measure like Mean Absolute Deviation (MAD) clarifies the scale of forecast errors.

Forecasting at Disney

• Disney's daily reports summarize yesterday's attendance forecasts and current attendance.
• Error measures, like MAPE, are expected to approach zero.
• An annual forecast for 2000 achieved a MAPE of 0.

Using The Computer to Forecast

• Computer programs, including specialized forecasting software, aid in handling intricate time series or causal models.
• These tools automate the selection of optimal model parameters.
• Integrating the process within inventory planning and control systems improves efficiency.

Test your understanding of forecasting methods with this quiz focused on alpha values in quarterly forecasting. You'll answer questions regarding tonnage predictions, absolute deviation, and comparisons of different alpha rates for better accuracy in forecasts.