Chapter 3 forecasting PDF

Preview

Summary

Professor Christopher Brock, contains bulleted list of important topics covered in chapter 3 of the textbook....

Description

BUSI 411 Chapter 3: Forecasting forecast – estimate about the future value of a variable such as demand Forecasts provide information on future demand The primary goal of operations management is to match supply to demand Businesses make plans for future operations based on anticipated future demand which can be derived from two sources::: actual customer orders and forecasts. Two important aspects of forecasts – 1) expected level of demand. 2) degree of accuracy that can be assigned to the forecast Forecasts are made with reference to a specific time horizon and can be short-term (pertains to ongoing operations) or long-term (pertain to new or long items) Two uses for forecasts:::: 1) help managers plan the system. 2) help managers plan the use of the system Forecasting techniques generally assume that the same underlying causal system that existed in the past will continue to exist in the future Important notes about forecasting::: 1) forecasts are not perfect. 2) forecasts for groups of items tend to be more accurate than forecasts for individual items. 3) forecast accuracy decreases as the time period covered by the forecast increases. A properly prepared forecast should fulfill certain requirements::: 1) timely (forecasting horizon must cover the time necessary to implement possible changes). 2) accurate (degree of accuracy should be stated). 3) reliable (work consistently). 4) meaningful units (the choice of unit should reflect user needs). 5) in writing (a written forecast will allow an objective basis for evaluation once actual results are in). 6) simple to understand and use. 7) cost-effective (benefits should outweigh the costs). Six steps in the forecasting process:::: 1) determine the purpose of the forecast. 2) establish a time horizon. 3) obtain, clean, and analyze appropriate data. 4) select a forecasting technique. 5) make the forecast. 6) monitor the forecast errors. Two general approaches to forecasting = qualitative and quantitative Judgmental forecasts – rely on analysis of subjective inputs obtained from various sources (ex. consumer surveys, sales staff, experts). Time-series forecasts – attempt to project past experience into the future (use historical data with the assumption that the future will follow the past) Associative models – use equations that consist of one or more explanatory variables that can be used to predict demand (ex. price per gallon, $ spent on ads) Qualitative forecast models = based on executive opinions (small group of upper-level managers that can meet and develop a forecast), salesforce opinions (members of the sales staff or customer service staff), consumer surveys (sampling of consumer opinions), or other approaches (delphi method [questionnaires to different people groups]) Time-series forecast models = an analyst must understand the underlying behavior of the time series which can be described as::: 1) trend (long-term upward/downward

movement in the data). 2) seasonality (short-term, fairly regular variations). 3) cycles (wavelike variations of more than one years duration). 4) irregular variations (unusual circumstances). 5) random variations (residual variations). Naïve forecast – uses a single previous value of a time series as the basis of a forecast (can be used with stable series, seasonal variations, or with trend) Averaging technique – smooth fluctuations in a time-series because the individual highs and lows in the data offset each other when combined. Three main averaging techniques:::: 1) moving average (uses a number of the most recent actual data values in generating a forecast). 2) weighted average (assigns more weight to the most recent values in a time series). 3) exponential smoothing Forecast (moving average method) = (however many ‘periods’ agreed upon [ex. 3-period uses data from last 3 years] added together) / number of periods Forecast (exponential smoothing method) = forecast for previous period + (smoothing constant % * (actual demand/sales for previous period – forecast for previous period)) Focus forecasting – uses several forecasting methods, all being applied to the last few months of historical data after any irregular variations have been removed Diffusion models – forecast based on rates of product adoption and usage spread from other established products Linear trend equation = (forecast for period = value of forecast at t0 + (slope of the line * specified number of time periods from t0)) Trend-adjusted exponential smoothing (double smoothing) – appropriate only when data vary around an average or have step or gradual changes (**see page 92 for formula**) Seasonality models = two main models include the additive model (demand = trend + seasonality) and the multiplicative model (demand = trend * seasonality). Seasonal relatives can be computed using the centered moving average or the simple average methods. Explanatory technique (for cycles) = searches for another variable that relates to, and leads, the variable of interest Associative forecasting models = rely on identification of related variables that can be used to predict values of the variable of interest Predicted (dependent variable) = value of dependent variable when independent variable is 0 + (slope of the line * independent variable) Standard error of estimate = square root of… ((sum of (values of each data point ^ 2)) / (number of data points – 2)) Linear regression (see *above formula) = one indicator of accuracy is the amount of scatter of the data points around the line 3 conditions are required for an indicator to be valid::: 1) relationship between movements of an indicator vs variable should have a logical explanation. 2) movements

of the indicator must precede movements of the dependent variable in enough time for the forecast to be helpful. 3) a fairly high correlation should exist between the 2 variables Correlation – measures the strength and direction of relationship between two variables use of simple regression analysis implies that::: variations around the line are random, deviations around the avg value are normally distributed, predictions are being made only within the range of observed values. To obtain the best results when using regression analysis::: always plot data, understand that the data may be time-dependent, know that a small correlation may imply that other variables are important. Decision makers will want to include accuracy as a factor when choosing among different forecasting techniques, along with cost. Forecast error = actual value – forecasted value Three commonly used measures for summarizing historical errors are the mean absolute deviation (MAD), the mean squared error (MSE), and the mean absolute percent error (MAPE) MAD = (sum of the absolute value of forecast errors (see *above error formula) for a given time period) / time period MSE = (sum of the (forecast errors for a given time period ^2)) / time period – 1 MAPE = (((sum of the absolute value of forecast errors for a given time period) / actual value) * 100) / time period Sources of forecast errors::: 1) model may be inadequate. 2) irregular variations may occur. 3) random variations Control chart = errors are plotted on a control chart in the order that they occur, center line of the chart represents an error of 0 to construct a control chart::: 1) compute the MSE. 2) figure out the control limits. Control limits = 0 + or – number of standard deviations from the mean (square root of the MSE) Tracking signal = (sum of the periods errors (see *above error formula)) / MAD for the period the control limits approach is far superior to the tracking signal approach the two MOST important factors when choosing a forecasting technique are COST and ACCURACY ((generally, the higher the accuracy the higher the cost)). However, other factors to consider are the availability of historical data, the availability of computer software, and the time constraints. Better short-term forecasts will not only enhance profits through lower inventory levels, fewer shortages, and improved customer service but they will also enhance forecasting credibility throughout the organization Maintaining accurate, up-to-date information on prices, demand, and other variables can have a significant impact on forecast accuracy

Companies may attempt to shorten the time horizon that forecasts must cover (because short-term forecasts have greater accuracy) by shortening lead time, building flexibility, or shortening the time needed to develop new products and services...