CHAPTER 3: FORECASTING (OPMAN W/ TQM) EXAMPLES OF USES OF FORECAST IN BUS. ORG.

:  RELIABLE – it should work consistently

 Expressed in MEANINGFUL UNITS – the
FORECAST ACCOUNTING, FINANCE, MARKETING, MIS, choice of units depends on user needs.
OPERATIONS, PRODUCT/SERVICE DESIGN  Be IN WRITING
An estimate about the future value of a variable of
 SIMPLE TO UNDERSTAND AND USE
interest such as demand.
2 USES OF FORECASTS: Several forecasting techniques: SIMPLE AND
A basic input in the decision process of operations COMPLEX
 PLAN THE SYSTEM – planning the system  COST-EFECTIVE – the benefits should
management because they provide information on
involves long range plans outweigh the costs
future demand.
 PLAN THE USE OF THE SYSTEM – short range
The primary goal of operations management is to and intermediate-range planning which
match supply to demand. involve tasks such as planning inventory and FORECASTING & THE SUPPLY CHAIN
workforce levels.
ANTICIPATED DEMAND is derived from two possible ACCURATE FORECAST is very important for the supply
sources (ACTUAL CUSTOMER ORDER & FORECASTS). FORECASTING is all about predicting customer chain. INACCURATE FORECASTS can lead to shortages
DEMAND. and excesses throughout the supply chain.
2 aspects of forecast: EXPECTED LEVEL OF DEMAND &
DEGREE OF ACCURACY FEATURES COMMON TO ALL FORECASTS STEPS IN THE FORECASTING TECHNIQUES

The EXPECTED LEVEL OF DEMAND can be a function of FORECASTING TECHNIQUES generally assume that the 1. Determine the purpose of the forecast. This
some structural variation, such as a trend or seasonal same underlying causal system that existed in the past step is an indication of the level of detail
variation. will continue to exist in the future. required in the forecast, the amt of
resources that can be justified and the level
FORECAST ACCURACY - a function of the ability of  Forecasts are not perfect; actual results of accuracy necessary.
forecasts to correctly model demand, random variation usually differ from predicted values; the 2. Establish a time horizon. The forecast must
and sometimes unforeseen events. presence of randomness precludes a perfect indicate a time interval, keeping in mind that
forecast. (Forecast is not perfect yet the accuracy decreases as the time horizon
Forecast are made w/ reference to a specific time level of accuracy is high somehow) increases.
horizon.  Forecasts for groups of items tend to be 3. Obtain, clean, and analyze appropriate data.
more accurate than forecasts for individual Once obtained, the data may need to be
SHORT-TERM FORECASTS – pertain to ongoing
items – cancelling effect cleaned to get rid of outliers and obviously
operations.
 Forecast accuracy decreases as the time incorrect data before analysis.
LONG-TERM FORECASTS – pertain to new products, period covered by the forecast- the time 4. Select a forecasting technique
new equipment that will require somewhat long lead horizon -increases. Forecast for short-term is 5. Make the forecast
time to develop, construct, or otherwise implement. more reliable. 6. Monitor the forecast errors. This is to know
if the objective is being met.
Forecast are the basis for budgeting, planning capacity, ELEMENTS OF A GOOD FORECAST [TARMISiCoCo]
sales, production and inventory, personnel, purchasing, APPROACHES TO FORECASTING
 A forecast should be TIMELY – data or
and more. Forecasts play an important role in the
evidence should be latest & up to date 2 General Approaches to Forecasting:
planning process because they enable managers to
 ACCURATE – degree of accuracy should be
anticipate the future so they can plan accordingly. QUALITATIVE & QUANTITATIVE
stated as this will enable users to plan for
possible errors
QUALITATIVE METHODS - MOST LIKELY – the estimate level of the  IRREGULAR VARIATION – caused by unusual
expected demand from salesforce which will circumstances, not reflective of typical
- Consist mainly of subjective inputs, which have the highest chance of occurrence. behavior. (weather conditions, strikes, or a
often defy precise numerical description. major change in product or service)
- Soft information (human factors, personal  CONSUMER SURVEYS  RANDOM VARIATION – residua; variations
opinion, hunches) - A considerable amt of knowledge & skill is that remain after all other behaviors have
required to construct a survey and correctly been accounted for.
QUANTITATIVE METHODS
interpret the results for valid info.
- Time-consuming, expensive, NAIVE METHODS
- Projection of historical data
- Consist mainly of analyzing objective, or hard A forecast for any period that EQUALS the previous
data  OTHER APPROACHES
period’s actual value
DELPHI METHOD
VARIETY OF FORECASTING TECHNIQUES: - Iterative process intended to achieve a A simple but widely used approach to forecasting
consensus forecast.
 JUDGEMENTAL FORECASTS TECHNIQUES FOR AVERAGING
- Group of experts
- Subjective inputs such as opinions from
- Involves circulating a series of
consumer surveys, sales staff, managers, AVERAGING TECHNIQUES generate forecasts that
questionnaires among individuals who
executives, & experts reflect recent value of a time series (the average value
possess the knowledge & ability to
over the last several periods)
contribute meaningfully.
 TIME-SERIES FORECASTS
- Useful for technological forecasting; for 3 TECHNIQUES FOR AVERAGING:
- Project patterns identified in recent time-
assessing changes on technology & their
series observations.
impact on an org.  MOVING AVERAGE
- Technique that averages a number of recent
 ASSOCIATIVE MODEL FORECASTS BASED ON TIME-SERIES DATA actual values, updated as new values
- Uses explanatory variables for predicting
become available.
future demand.  TIME SERIES
- ADVANTAGE: EASY TO COMPUTE & EASY TO
- A time ordered sequence of observations
UNDERSTAND
taken at regular intervals
- DISADVANTAGE: ALL VALUES IN THE
- A set of numerical measurements that is
QUALITATIVE FORECASTS AVERAGE ARE WEIGHTED EQUALLY
taken equally spaced intervals over time
 EXECUTIVE OPINIONS - Analysis of TS requires the analyst to identify
 WEIGHTED MOVING AVERAGE
- Small group of upper-level managers the underlying behavior of the series which
- More recent values in a series are given
(marketing, operations, & finance) can be accomplished by plotting the data.
more weight in computing a forecast
- Used as part of long-range planning and new - ONE OR MORE PATTERNS MIGHT APPEAR:
product development.  TREND – a long term upward or downward
 EXPONENTIAL SMOOTHING
movement in data
- A weighted averaging method based on the
 SALESFORCE OPINIONS  SEASONALITY- short-term regular variations
previous forecast plus a percentage of the
- Lower positions related to the calendar or time of day
forecast error.
- Have direct contact w/ customers  CYCLE – wavelike variation lasting more than
- 3 levels: OPTIMISTIC, MOST LIKELY, 1 yr. (variety of economic, political, &
PESSIMISTIC agricultural conditions)
OTHER FORECASTING METHODS REGERESSION – technique for fitting a line to a set of 2. Irregular variations may occur due to severe
points weather or other natural phenomena,
 FOCUS FORECASTING temporary shortages or breakdowns,
- Based on a “best recent performance” basis. LEAST SQUARES LINE – minimizes the sum of the catastrophes, or similar events
BERNARD T. SMITH squared vertical deviations around the line 3. Random variations. Randomness is the
 DIFFUSION MODELS inherent variation that remains in the data
- When new products or services are STANDARD ERROR OF ESTIMATE – a measure of the
after all causes of variation have been
introduced, historical data are not generally scatter of points around a regression line
accounted for. There are always random
available on which to base forecasts. variations.
CORRELATION – measure of strength and direction of
- Widely used in marketing and to assess the
relationship between 2 variables.
merits of investing in new technology. CONTROL CHART – visual tool for monitoring forecast
errors
TECHNIQUES FOR TREND
FORECAST ACCURACY TRACKING SIGNAL – the ratio of cumulative forecast
 TREND EQUATION error corresponding value of MAD, used to monitor a
- Linear trend equation is used to develop Accuracy and control of forecast is a vital aspect of forecast
forecasts when trend is present. forecasting, so forecasters want to minimize errors.
 EXTENSION OF EXPONENTIAL SMOOTHING BIAS – persistent tendency for forecast to be greater or
ERROR – difference between the actual value and the less than the actual values of a time series
TREND-ADJUSTED EXPONENTIAL SMOOTHING value that was predicted for a given period

- Variation of exponential smoothing used SUMMARIZING FORECAST ACCURACY

when a time series exhibits a linear trend.
- Also known as DOUBLE SMOOTHING  MAD/MEAN ADSOLUTE DEVIATION – the
average absolute forecast error
 MSE/MEAN SQUARED ERROR – the average
SEASONALITY – regular annual variations of squared error
 MAPE/MEAN ABSOLUTE PERCENT ERROR –
SEASONAL VARIATIONS - Regularly repeating
the average absolute percent error
movements in series values that can be tied to
recurring events
MONITORING FORECAST ERROR
SEASONAL RELATIVE – percentage of average or trend
There are a variety of possible sources of forecast
CENTERED MOVING AVERAGE – a moving average
errors:
positioned at the center of the data that were used to
compute it 1. The model may be inadequate due to:
- The omission of an important variable
- A change or shift in the variable that the
ASSOCIATIVE FORECASTING TECHNIQUES model cannot deal with (the sudden
appearance of a trend or cycle)
PREDICTOR VARIABLES – variables that can be used to - The appearance of a new variable (new
predict values of the variable of interest competitor)