270 likes | 478 Views
BA339 – OM, Chapter 4. Forecasting What is it? Definition & Time Horizons Type of Forecasts Strategic Importance Seven Steps of Forecasting Forecasting Approaches – Qualitative & Quantitative Designing the Forecasting System Review of Quantitative Forecasting Approaches.
E N D
BA339 – OM, Chapter 4 • Forecasting • What is it? Definition & Time Horizons • Type of Forecasts • Strategic Importance • Seven Steps of Forecasting • Forecasting Approaches – Qualitative & Quantitative • Designing the Forecasting System • Review of Quantitative Forecasting Approaches
BA339 – OM, Chapter 4 • Forecasting Defined • “Art & science of predicting future events for planning purposes” • Can be subjective or intuitive • Can use historical data and mathematical modeling • Used to: • Determine resources needed, schedule existing resources & acquire additional resources • Anticipate changes in prices/costs • Prepare for new laws/regulations, competitors, resource shortages or technologies • Generally driven by customer demand
BA339 – OM, Chapter 4 • Forecasting Time Horizons • Short-range: • Generally < 3 mos., but can be up to one yr. • Used for planning purchases, job scheduling, workforce levels, job assignments & production levels • Medium-range • Generally 3 mos. To 3 yrs. • Useful in sales planning, production planning & budgeting, cash management & analyzing operating plans • Long-range • Generally > 3 yrs. • Used in planning new products, capital expenditures, facility location/expansion & R&D
BA339 – OM, Chapter 4 • Medium- & long-range (M&LR) vs. Short-range (SR) Forecasts • M&LR deal with more comprehensive issues: planning & products, plants, & processes (Ex. – GM manufacturing plant in Brazil) • SR employs different methodologies: mathematical techniques (moving avg., smoothing, trend extrapolation) more common to SR; M&LR use broader, less quantitative approaches • SR forecasts tend to be more accurate: factors influencing demand change daily, thereby diminishing accuracy over time • All are affected by product lifecycle – products & services do not sell at constant levels throughout the lifecycle • Products in the introduction & growth stage need longer forecasts than products in maturity & decline stage
BA339 – OM, Chapter 4 • Types of Forecasts • Economic – address business cycles by predicting inflation rates, money supply, housing starts, etc. • Example – ISM’s Report on Business • Technological – deal with rates of technological progress which affects new product development that may require new plants and equipment • Demand – projections of demand for products & services. • Also called sales forecasts • Drive production, capacity, purchasing & scheduling systems and serves as inputs to financial, marketing, and personnel planning • Essentially a process of aggregating information
BA339 – OM, Chapter 4 • Patterns of Demand • Repeated observations for a product over time form a pattern know as demand time series > 5 basic patterns • Horizontal – fluctuation of data around a constant mean • Trend – systematic increase/decrease in the mean of the series over time • Seasonal – repeatable pattern of increase/decrease in demand depending on the time of day, week, month, or season • Cyclical – less predictable gradual increases/decreases in demand over longer periods of time (years/decades) • Random – unforecastable variation in demand
BA339 – OM, Chapter 4 • Factors Affecting Demand • External • Expansion vs. contraction of economy; lack of uniformity • Regulation – Ex., Limiting the sulfur content of coal will reduce demand for high sulfur, increase low sulfur demand and not affect demand for electricity • Economic time series forecasting (government and private) not an exact science but is useful in estimating changes • Turning Point is critical – point where long-term rate of growth in demand will change
BA339 – OM, Chapter 4 • Factors Affecting Demand - External • Leading Indicators • External factors w/ turning points that typically precede peaks & troughs of business cycle • Examine rate of change vs. frequency of change • Ex. – Residential building contracts might precede plywood sales by several weeks, homeowner’s insurance by several months, and furniture sales by one year • Coincident indicators • Time series with turning points that generally match those of the general business cycles (Ex., employment figures) • Lagging indicators • Follow turning points, typically by several weeks or months (Ex., retail sales)
BA339 – OM, Chapter 4 • Factors Affecting Demand • Internal • Decisions about product/service design, price, advertising promotions, packaging design, sales incentives, etc. influence demand volume • Demand management – describes the process of influencing the timing/volume of demand or adapting to undesirable effects of unchangeable demand patterns. Ex. – auto makers use of rebates to boost car sales post 9/11 • Timing/influencing of demand is a critical factor in efficiently utilizing resources and production capacity • Ex. – Telecomm encourages evening/weekend long distance to spread demand more evenly • Ex. – Engine mfg. for lawn tractors make make engines for snowmobiles/snow blowers to even out production/resource requirements • Ex. – Doctors/dentists use appointments to control demand
BA339 – OM, Chapter 4 • Strategic Importance of Forecasting • Forecast is the only estimate of demand until actual demand becomes known • Core piece of operations; permits proactive planning vs. reactive • Drives decision-making in several areas: • HR – hiring, training, laying off workers driven by demand for goods & services. Sudden changes can affect product quality, customer satisfaction, etc. • Capacity • Shortages – loss of customers, market share • Excess – negatively affect cash flow, investment, profitability • Supply Chain Management – accurate forecasts support good supplier relations, capital investment, profit margins
BA339 – OM, Chapter 4 • Seven Steps of Forecasting • Determine the use of the forecasting • Select the items to be forecasted • Level of aggregation - increased accuracy • Units of measure – number of cars vs. total sales • Determine the time horizon of the forecast • Short-, medium-, or long-range • Select the type of forecasting technique • Qualitative vs. quantitative • Gather the data needed to make the forecast • Make the forecast • Validate and implement the results • Check assumptions with affected departments (stakeholders) • Use forecast to schedule materials, etc.
Application Short-term (0-3 months) Medium-term (3 mos. – 2 yrs.) Long-term more than 2 yrs.) Forecast Quantity ØIndividual products or services ØTotal sales ØGroups of families of products or services ØTotal sales Decision Area ØInventory Mgmt. ØFinal Assembly Scheduling ØWorkforce Scheduling ØMaster Production Scheduling ØStaff Planning ØProduction Planning ØMaster production Scheduling ØPurchasing ØDistribution & Logistics ØFacility Location ØCapacity Planning ØProcess Management Forecasting Technique ØTime series ØCausal ØJudgment ØCausal ØJudgment ØCausal ØJudgment BA339 – OM, Chapter 4
BA339 – OM, Chapter 4 • Forecasting Approaches • Qualitative • Incorporates intuition, emotion, personal experiences, & value systems • Used when adequate historical data is lacking • Relies on managerial judgment/experience • Can be used to modify forecasts generated by quantitative methods • Includes: • Executive opinion • Sales force estimates • Market research/survey • Delphi method
BA339 – OM, Chapter 4 • Qualitative Forecasts • Executive Opinion • Uses opinions of one or more managers/technical experts, often in combination with statistical data to estimate demand • Advantages – Can be relatively quick, take advantage of broad experience of multiple managers • Disadvantages – Tendency toward “group-think”, can be costly (demand on managers’/executives’ time, failure to document assumptions & track changes limits accuracy and utility • Examples: Tupperware’s consensus from sales, marketing, finance, & production; Bristol-Meyers Squibb use of 220 established research scientists to grasp future trends.
BA339 – OM, Chapter 4 • Qualitative Forecasts • Sales Force Estimates • Individual salesperson projections aggregated at district & national levels, then reviewed to ensure overall accuracy/realism • Advantages – Sales force most likely to know customer product desires, future needs, & quantities; regional projections helpful in inventory/distribution management and staffing; facilitates aggregation of data • Disadvantages – Individual biases (up & down); problems with determining customer needs vs. wants; implicit incentive for over- or under-estimating (depending on incentives or minimum sales
BA339 – OM, Chapter 4 • Qualitative Forecasts • Market Research/Survey • Solicits input from customer/potential customers regarding future purchasing plans; can be relatively unstructured to very systematic • Advantages – Can be helpful in improving product design and planning new products; good short-term accuracy • Disadvantages – numerous qualification/hedges dilutes accuracy; poor response rate to questionnaires limits validity; imitative nature of some questionnaires limits utility due to limited customer reference
BA339 – OM, Chapter 4 • Qualitative Forecasts • Delphi Method • Iterative group process w/ input from decision makers, staff, & respondents; coordinator consolidates input; anonymous input valuable to ensure independence, avoid expert dominance • Advantages – Iterative process helps reach consensus, reduces “group-think”, permits experts to devote attention to following scientific advances, govt regulations, & competitive environment • Disadvantages – time-consuming, response less meaningful due to lack of accountability of respondents, poor questionnaires/data-gathering leads to false conclusions
BA339 – OM, Chapter 4 • Qualitative Forecasts • Guideline for Use • Adjust quantitative forecasts when their track record is poor and the decision maker(s) have important contextual knowledge • Ensure that contextual information is identified by all personnel providing input • Make adjustments to quantitative forecasts to compensate for specific events • Example – advertising campaigns, competitors actions, or international developments
BA339 – OM, Chapter 4 • Quantitative Forecasts – Time Series & Associative • Time-series • Sequence of evenly spaced data points gathered at regular time periods (days, wks….) • Based on past values • Four components: • Trend • Seasonality • Cycles • Random variations • Methods • Naïve approach • Moving average • Exponential smoothing • Trend projection/analysis
BA339 – OM, Chapter 4 • Time-Series Forecasting Models • Naïve Forecast • Forecast of the next period equals the current period; forecast of the same time last year reflects current time this year (Ex., July > July) • Provides a starting point for more sophisticated models • Can be used to create a simple demand trend (Ex., rate of change (increase/decrease) between last two periods used to project next month demand) • Advantages – simplicity, low cost; works best when horizontal, trend, or seasonal patterns are stable and random variation is small • Disadvantages – substantial impact in accuracy if random variation is significant; limited planning horizon
BA339 – OM, Chapter 4 • Time-Series Forecasting Models • Moving Average • Uses a number of historical actual data values; calculate mean of the demand for identified periods; can involve weighting various periods • Simple Moving Average = Sum of last n demands/n • Weighted Moving Average = Sum (weight for period n)(demand in period n)/Sum of weights • Permits you to emphasize recent demand over earlier demand; more response than simple moving average to changes in underlying avg. of demand series • Increasing n smoothes out fluctuations better but is less sensitive to real changes in data; does not pick up trends very well; do not predict changes to either higher or lower levels (stay within past levels) – lag the actual values; require extensive records of past data
BA339 – OM, Chapter 4 • Time-Series Forecasting Models • Exponential Smoothing • Sophisticated weighted moving-average method that calculates the average of a time series by giving recent demands more weight than earlier demand • Requires only: Last period forecast, demand for this period and a smoothing parameter (alpha) • New forecast = last period’s forecast +alpha(last period’s actual demand – last period’s forecast) • Example – P.87 of text: 142 + .2(153-142) = 142 + 2.2 = 144.2 • Larger alpha values emphasize recent demand levels and are more responsive to to changes in underlying avg.; lower value emphasize past data • Advantages – simple, minimal data requirements • Alpha values generally range from .05 to .50 for business applications
BA339 – OM, Chapter 4 • Time-Series Forecasting Models • Trend Projections • Fits a trend line to a series of historical data points and projects the line into the future for medium- to long-range forecasts • Several trend equations – exponential, quadratic, and linear • Least Squares Method – see pages 93-95 (test) • Minimizes the sum of squares of the deviations from the line to each of the actual observations; line is described in terms of y-intercepts and its slope • Equation: y = a + bx; • Notes: Always plot data since least squares assumes a linear relationship; do not predict time periods far beyond given database; deviations are assumed to be random and normally distributed
BA339 – OM, Chapter 4 • Time-Series Forecasting Models • Minimizing Forecasting Error • 2 measures – measure the dispersion of forecast errors; larger the value the larger the forecast error • Mean Absolute Deviation = Sum of absolute values of individual forecast errors / number of periods of data • Mean Square Error = Average of the squared differences between the forecasted and observed values
BA339 – OM, Chapter 4 • Associative Forecasting Methods • Usually consider several variable that are related to the quantity being predicted; once the related variable are found, statistical models are then built and used to forecast • Permits consideration of many variables • Example: PC sales forecasts (dependent variable) could be correlated to advertising budget, promotions, prices, competitors prices (independent variables) • Linear-regression analysis • Utilizes least squares method of trend projection, but independent variable, x, will no longer be time. • Requires accuracy measurement of regression estimates > Standard error of estimate • Correlation coefficient – measures relationship between two variables (degree or strength of linear relationship)
BA339 – OM, Chapter 4 • Monitoring & Controlling Forecasts • Tracking Signal - TS • Used to determine accuracy of forecast – e.g., how well does the forecast predict the actual values • TS = RSFE/MAD • RSFE (Running Sum of Forecast Errors) • MAD – Mean Absolute Deviation • Positive TS means demand is > forecast • Negative TS means demand is < forecast • Once calculated, TSs are compared to pre-established upper and lower control limits (reflects application of quality principles to forecasting)
BA339 – OM, Chapter 4 • Forecasting in the Service Sector • Characterized by unusual demand patterns depending on the service • Emphasis on maintaining good short-term records • More art than science; less emphasis on complex statistical analysis • More frequent use of point-of-sale computers to track sales by time period • Example: Restaurant tracking of sales by hour, half-hour, or quarter-hour