Operations Management and TQM PDF

Operations Management and TQM BME 3003 / Midterms / SC 529 This highlights the fact that customers' Chapter 7: Forecasting and Demand Planning wants and needs d...

Operations Management and TQM BME 3003 / Midterms / SC 529 This highlights the fact that customers' Chapter 7: Forecasting and Demand Planning wants and needs define the customer Forecasting – is the process of projecting the benefit package, and that customer values of one or more variables into the future. demand pulls goods and services through the supply chain. Good Forecasts – are needed in all organizations to drive analyses and Forecast Planning Horizon decisions related to operations. Planning Horizon – is the length of time on Poor Forecasting – can result in poor which a forecast is based. inventory and staffing decisions, resulting Long-range Forecasts - cover a planning in part shortages, inadequate customer horizon of 1 to 10 years and are service, and many customer complaints. necessary to plan for the expansion of facilities and to determine future needs for MANY FIRMS INTEGRATE FORECASTING land, labor, and equipment. WITH VALUE CHAIN AND CAPACITY MANAGEMENT SYSTEMS TO MAKE BETTER Intermediate-range Forecasts - over a 3- OPERATIONAL DECISIONS. to 12-month period are needed to plan workforce levels, allocate budgets among divisions, schedule jobs and resources, Organizations make many different types and establish purchasing plans. of forecasts. Short-range Forecasts - focus on the Top managers need long-range forecasts planning horizon of up to three months expressed in total sales dollars for use in and are used by operations managers to financial planning and for sizing and plan production schedules and assign locating new facilities. workers to jobs, determine short-term capacity requirements, and aid shipping At lower organizational levels, however, departments in planning transportation managers of the various product groups needs and establishing delivery need aggregate forecasts of sales volume schedules. for their products in units that are more meaningful to them. FORECASTS OF FUTURE DEMAND ARE Some software vendors are beginning to NEEDED AT ALL LEVELS OF use the terms demand planning or demand ORGANIZATIONAL DECISION MAKING. chain instead of supply chain. Time Bucket - is the unit of measure for the time but similar repeatable patterns period used in a forecast. might occur over the weeks during a month, over days during a week, A time bucket might be a year, quarter, or hours during a day. month, week, day, hour, or even a minute. For a long-term planning horizon, a firm 3. Cyclical Patterns - are regular patterns in might forecast in yearly time buckets; a data series that take place over long periods of time. For a short-range planning horizon, the time bucket might be an hour or less. 4. Random Variation - (sometimes called noise) is the unexplained deviation Data Patterns in Time Series of a time series from a predictable pattern such as a trend, seasonal, or cyclical Statistical methods of forecasting are based on pattern. the analysis of historical data, called a time series. Random variation is caused by Time Series - is a set of observations short-term, unanticipated, and measured at successive points in time or nonrecurring factors and is over successive periods of time. unpredictable. A time series provides the data for Because of random variation, understanding how the variable that we forecasts are never 100 percent wish to forecast has changed historically. accurate. Five characteristics of time series: SELECTING THE RIGHT PLANNING HORIZON 1. Trend - is the underlying pattern of growth LENGTH AND TIME BUCKET SIZE FOR THE or decline in a time series. RIGHT SITUATION IS AN IMPORTANT PART Trends can be increasing or OF FORECASTING. decreasing and can be linear or nonlinear. 5. Irregular Variation – is a one-time variation that is explainable. 2. Seasonal Patterns - are characterized by To develop a reliable forecast for repeatable periods of ups and downs over the future, we would need to take short periods of time. into account both the long-term Seasonal patterns may occur over trend and the annual seasonal a year; pattern. We generally think of seasonal patterns occurring within one year;  Forecast Errors and Accuracy The values of MAD and MSE depend on All forecasts are subject to error, and the measurement scale of the time-series understanding the nature and size of errors is data. important to making good decisions. On the other hand, a variable like market Forecast Error - is the difference between share, which is measured as a fraction, the observed value of the time series and will always have small values of MAD and the forecast, or 𝐴𝑡 − 𝐹𝑡. MSE. Thus, the measures have no meaning Three types of forecast error metrics: except in comparison with other models 1. Mean Square Error or MSE - is used to forecast the same data. calculated by squaring the individual forecast errors and then averaging the MAPE is different in that the measurement results over all T periods of data in the scale factor is eliminated by dividing the time series. absolute error by the time-series data MSE is probably the most value. This makes the measure easier to commonly used measure of interpret. forecast accuracy. Statistical Forecasting Models Generally, the most popular Forecasting methods can be classified as either statistical or judgmental. Sometimes the square root of MSE is computed; this is called the root Statistical Forecasting - is based on the mean square error, RMSE. assumption that the future will be an extrapolation of the past. 2. Mean Absolute Deviation or MAD - This measure is simply the average of the sum Statistical methods can generally be of the absolute deviations for all the categorized as: forecast errors. time-series methods, which extrapolate historical time-series data, and 3. Mean Absolute Percentage Error or MAPE - This is simply the average of the regression methods, which extrapolate percentage error for each forecast value in historical time-series data but can also the time series. include other potentially causal factors that A major difference between MSE and influence the behavior of the time series. MAD is that MSE is influenced much more by large forecast errors than by small errors (because the errors are squared)  Simple Moving Average Regression as a Forecasting Approach The simple moving average concept is based on Regression Analysis - is a method for building a the idea of averaging random fluctuations in a statistical model that defines a relationship time series to identify the underlying direction in between a single dependent variable and one or which the time series is changing. more independent variables, all of which are numerical. Moving Average (MA) Forecast - is an average of the most recent “K” Causal Forecasting Models with observations in a time series. Multiple Regression MA methods work best for short planning In more advanced forecasting applications, other horizons when there is no major trend, independent variables such as economic indexes seasonal, or business cycle patterns-that or demographic factors that may influence the is, when demand is relatively stable and time series can be incorporated into a regression consistent. model. Multiple Linear Regression Model - A As the value of k increases, the forecast linear regression model with more than reacts slowly to recent changes in the time one independent variable series because older data are included in the computation. MULTIPLE REGRESSION PROVIDES A As the value of k decreases, the forecast TECHNIQUE FOR BUILDING FORECASTING reacts more quickly. MODELS THAT NOT ONLY INCORPORATE TIME BUT OTHER POTENTIAL CAUSAL If a significant trend exists in the time- VARIABLES. series data, moving-average-based forecasts will lag actual demand, resulting in a bias in the forecast. Judgmental Forecasting Judgmental Forecasting - relies upon opinions Single Exponential Smoothing and expertise of people in developing forecasts. Single Exponential Smoothing (SES) - is a Judgmental forecasting is possible. But forecasting technique that uses a weighted even when historical data are available average of past time-series values to forecast the and appropriate, they cannot be the sole value of the time series in the next period. basis for prediction. SES forecasts are based on averages using and weighting the most recent The demand for goods and services is actual demand more than older demand affected by a variety of factors such as data. global markets and cultures, interest rates, disposable income, inflation, and technology.  The level of aggregation often dictates the Competitors' actions and government appropriate method. regulations also have an impact. Forecasting the total amount of soap to produce over the next planning period is Several approaches are used in judgmental certainly different from forecasting the forecasts: amount of each individual product to Grassroots Forecasting - is asking those produce. who are close to the end consumer, such Aggregate forecasts are generally much as salespeople, about the customers' easier to develop, whereas detailed purchasing plans. forecasts require more time and resources. Delphi Method - consists of forecasting by expert opinion by gathering judgments The choice of forecasting method depends and opinions of key personnel based on on other criteria as well. their experience and knowledge of the situation. The time span is one of the most critical criteria. Different techniques are Forecasting in Practice applicable for long-range, intermediate- range, and short-range forecasts. In practice, managers use a variety of judgmental and quantitative forecasting Also important is the frequency of techniques. updating that will be necessary. Many managers begin with a statistical Forecasters should also monitor a forecast forecast and adjust it to account for such to determine when it might be factors. advantageous to change or update the model. Others may develop independent judgmental and statistical forecasts and A tracking signal provides a then combine them, either objectively by method for doing this by averaging or in a subjective manner. quantifying bias - the tendency of forecasts to consistently be larger The first step in developing a practical or smaller than the actual values of forecast is to understand its purpose. the time series. For instance, if financial personnel need a sales forecast to determine capital Ten Practical Principles of Forecasting investment strategies, a long (two- to five- 1. Use quantitative rather than qualitative year) time horizon is necessary. methods. 2. Limit subjective adjustments of quantitative forecasts. 3. Adjust for events expected in the future. 4. Ask experts to justify their forecasts in writing. 5. Use structured procedures to integrate judgmental and quantitative methods. 6. Combine forecasts from approaches that differ. 7. If combining forecasts, begin with equal weights. 8. Compare past performance of various forecasting methods. 9. Seek feedback about forecasts. 10. Use multiple measures of forecast accuracy.