Forecasting -Supply

A prediction of future events used for planning purposes-predicting future demand. Important for process management
Click the card to flip 👆
1 / 76
Terms in this set (76)
A prediction of future events used for planning purposes-predicting future demand. Important for process management
Forecast:
Based on multiple types of data or analysis: mathematical models (quantitative data), expert opinion, historical data
Forecast:
Used for process management & supply chain management
Forecast:
It creates a schedule to produce as much as you need for that demand or to make sure that you have employees available if it's a service setting to satisfy that demand. (supply chain management) order components from a supplier, you need to know what demand is.
Forecast:
plot your demand over time
Time Series
Repeated observations of demand for a product/service in their order of appearance
Time series
can reveal patters in demand over time
Time series data
how many units do we sell every day. And you'll plot that on a graph, any patterns or trends in demand over time
Time Series data
5 basic patterns you can expect from time series data:
Horizontal, Trend, Seasonal, Cyclical and Random
Judgment methods, Casual methods and Time-Series Analysis
What technique should you use?
Fluctuation of data around a constant mean, demand is consistent n predictable . ex: flour , cooking n bakingHorizontalSystematic increase or decrease in the mean over time. -the average demand is actually changing in a predictable wayTrend:Repeatable pattern of increases or decreases in demand, depending on time of day,week,month,season -how demand changes at different times of the day, or different days of the week, or different months of the year.Seasonal:Less predictable gradual increases/decreases in demand over longer periods of time (years or decades) ex: fashionCyclicalUnforecastable variation in demand. There is no relationship between past demand and future demand.Random• Same resources • Different demand cycles ex: assiniboin parkComplementary Services:• services that have an opposite demand cycle from what you would see in the summer, by offering winter activities there.Complementary Services:• if they didn't have fun things to do with the park in the winter, then they would have a lot of resources that are sitting unused throughout the year. So all of the pavilions, the washrooms the places you can eat.Complementary Services:-Increase demand -Shift demand to new periodPromotional Pricing :shift demand forward by offering a sale to encourage people to buy more now, so you're not holding on to that inventory for as longPromotional PricingLevel demand based on capacityPrescheduled AppointmentsMaking an appointment with a doctor, or a dentist, or a hairstylist, or especially personal care services, they look at how many, you know, practitioners do we have available? And how do we ensure that they're used in a steady way throughout the day, and that we don't buildup of demand really, really early and lead to long customer waitsPrescheduled AppointmentsAdjust prices in real time based on demand.Revenue Management :Airline industry. Price increase, change price based on demand. Also shifting amount of seats.Revenue Management :- prices are adjusted so they can maximize the amount of revenue that they generate from the resources.Revenue Management :-Accumulate orders for future delivery• -Decrease service level and risk losing customersBacklogs, Backorders, Stockoutswe'll take your order now. But we aren't going to be able to satisfy it right away, youll get product in the future.Backlog, or backorderwhen they don't have the item, go buy somewhere else.stockoutyou're accumulating demand for the future when maybe you have more capacity to satisfy it.Backlogs, Backorders, StockoutsHistory of Past Demand, Notes Explaining Past Demand, Past Forecasts, Consumer Research, Planned Promotions, Inputs from Partners via CPFR1. What are your inputs?what has the observed demand been recentlyHistory of Past Demandthat horizontal pattern where we had that weird blip in the middle, maybe a celebrity tweeted something negative about your company. how useful that observation is in predicting future demand or whether it's something that should kind of be discarded.Notes Explaining Past Demandrather than what the observed demand was, you can look at, well, what did we expect demand to be. the observed demand can also tell you if the way that you've been generating forecasts in the past is working well.Past Forecastsyou contract another company out to do on your behalf, to get a sense of what consumers might be looking for. if you don't have data, you can start to figure out what would be appropriate for your company.Consumer ResearchIt would also be useful to know if your organization has any sales or promotions coming up. because they're likely to generate a big increase in demand that you'll want to account for it in your planning.Planned Promotionsinput that you can use would be feedback from your partners that you capture using collaborative planning, forecasting and replenishmentInputs from Partners via CPFRis a supply chain and forecasting practice, where every actor in the supply chain, so the Suppliers and distributors that retailers, they get together, and they each generate their own forecasts for what they expect demand to beCPFRCollaborative Planning, Forecasting &ReplenishmentCPFR-Requires collaboration with suppliers, customers -Independent forecasts generated & compared,adjusted until forecasts approach consensusCPFRComplementary Services, Promotional Pricing, Prescheduled Appointments, Revenue Management, Backlogs, Backorders, StockoutsManaging demandwhen you can combine the knowledge that everyone in the supply chain has, you're able to generate maybe a more accurate prediction of what demand is going to be. helps to build trust in the relationship, everyone's working together for kind of a shared benefit, which creates maybe a more collaborative environment in the supply chain.What are the benefits to suppliers? RetailersIndividual Products vs. Product Families, Units of Measurement: Units vs. $ vs. LabourWhat is it that you are trying to predict?-what level of aggregation are we trying to generate this forecast -the time horizon you're looking at and the level of aggregation are both important decisionsIndividual Products vs. Product FamiliesApple's coming up with a forecast. Depending on the time horizon, they might say, let's predict demand for iPhones as a whole, rather than predicting demand for each individual model of iPhone. And what that does aggregating at the level of a product family is that it helps smooth out variation.Individual Products vs. Product Familiesare you trying to predict how many units of a product we're going to sell? Or are you trying to predict revenue for the next quarter, for example, are you trying to predict how much labor or resources we're going to need to be able to satisfy demand, because each of those different units will require different inputs from other groups within the organization.Measurement: Units vs. $ vs. LabourJudgment methods, Casual methods and Time-Series AnalysisWhat technique should you use?qualitative. talking to experts talking to customer focus groupsJudgment methodspredictive modeling, you can use regression analysisCasual methodsassuming that past demand is a good predictor of future demand.Time-Series AnalysisSubjective Can incorporate a variety of information Do not require numerical data Results may be biased Results may be conflictingQUALITATIVEallows you to really benefit from the expertise of specific individuals.QUALITATIVE-Objective Can incorporate large a volume of information Do not have to rely on few individuals Numerical data may not be available Mathematical models may be too simplisticQUANTITATIVEas long as someone has the same data set, and they're working with the same forecasting model, they should be able to get the same outputs.QUANTITATIVEMakes predictions based only on historical data about the dependent variable.TIME SERIES METHODSAssumes that past demand is a useful predictor offuture demandTIME SERIES METHODS● Popular methods: 1. Naive 2. Simple moving average 3. Weighted moving average 4. Exponential smoothingTIME SERIES METHODSForecast for next period equals demand for most recently observed period ● Only appropriate for short-term forecasts ● Sensitive to random variationNaive MethodFt+1 = DtNaive Method= forecast for period t+1 (or for any future period)Ft+1=actual demand in the mostrecent period tDtForecast for next period equals average demand for n most recent periods ● Smooths out random variation. Each observation is equally weightedSimple Moving Average= total number of periods used in calculation ● Sensitivity to random variation depends on N and N is subjectiveNSo if you have a lot of variation in your demand day to day, you might want to use a larger end value. If demand is pretty consistent, it might be safer for you, for you to use a smaller n value,N as well● Forecast for next period equals average demandfor n most recent periods, and each observation of demand can have its own weight; the sum ofweights = 1 ● Weights represent the varying amounts ofinfluence of past demand on forecast. still have to choose that end valueWeighted Moving Average:weight assigned to t's demand -Weights and N are subjectiveWt● A weighted moving average assigning differing levels of weight to recent demand compared toolder historical data ● Requires only three data points: last period's forecast,last period's demand, smoothing parameter (α)Exponential Smoothing● smoothing parameter (0<a£1); commonly 0.01-0.5 ● Low a®more smoothing; high a®more responsive tochangesais subjective. Alpha determines again, how much weight you're drawing from the observed demand and how much weight is coming from the forecast.a as wellA seasonal factor is calculated, which is then multiplied by anestimate of average demand to adjust for seasonalityMultiplicative Seasonal MethodDifference between predicted demand and actual demand Et = Dt - Ft Can be random or the result of an inappropriate modelForecast Error1. Cumulative Sum of Forecast Errors(CFE) 2. Mean Squared Error (MSE) 3. Mean Absolute Deviation (MAD) MeanAbsolutePercentageDeviationMeasures of Forecast Error:Assesses total error in forecasts over time CFE = S EtCumulative Sum of Forecast Errors (CFE)● Used to evaluate presence and direction of bias ● If forecast is consistently lower than demand, CFE will behighly positive ● If forecast consistently higher than demand, CFE will behighly negative ● As close as zero as possibleCumulative Sum of Forecast Errors (CFE)● Indicates on average, how close forecast is to demand ● Magnifies large errors ● Standard deviation accomplishes same functionMean Squared Error (MSE) S Et2 / n● Simple measure of magnitude of error ● Does not reveal directional biasMean Absolute Deviation (MAD) S |Et| / nContextualizes magnitude of error relative to demandMean Absolute Percentage Error (MAPE)