Essay On Forecasting Flashcards

Unlock all answers in this set

Unlock answers
question
Forecasting Definition and Applications
answer
The art and science of predicting future events and outcomes Applications: -Economic Forecasts: Predicting changes of business cycles -Technological Forecasts: rates of technological change -Business Forecasts: predicting demand
question
Forecasting Means
answer
Anticipating & quantifying what needs to be done to meet the customer's needs/requirements Planning for the resources needed to meet future demand Arming management with the info needed to make the best possible business decisions (to minimize risks inherent in decision making)
question
Forecast: Basic Principles
answer
Forecasts are never accurate Forecasting techniques assume a causal relationship Forecasting aggregate units tend to give more accurate results than forecasting individual units Forecasting accuracy decreases as the time horizon increases **It's more than just predicting future demand
question
Forecasting Requirements:
answer
Organization Defining approach Methodology -Top-down or Bottom-Up -Short term vs. Long Term -Denationalized vs. Raw Data Simulating forecasting performance -Include all factors influencing demand -Historical Evaluation Monitoring accuracy Consistency Easy to Understand
question
Choosing a Forecast Model
answer
-Chose the forecast model that makes best use of available data -Consider the stage of the product's life cycle -Selecting applicable forecast models, use relevant or historical data -Simulating a forecast performance and determine which forecast model gives the best accuracy
question
Forecasting Error
answer
Difference between the actual demand ; its forecasted value *Perfect Forecast is when Actual = Forecast, which is impossible
question
Error Magnitude tools
answer
Mean Absolute Deviation (MAD) Mean Absolute Percentage Error (MAPE) Mean squared error (MSE)
question
Mean Absolute Deviation
answer
Measures absolute deviation of forecast from actual. *Positive ; negative errors do not cancel out *Goal: Seek MAD ratios to be as small as possible
question
Mean Absolute Percent Error
answer
The MAPE ratio is the same as the MAD ratio, except This measures deviation as a percentage of actual data
question
Mean Squared Error
answer
MSE measures squared forecast error **Recognizes that large errors are disproportionately more "expensive" than small errors
question
Cumulative Sum of Forecast Errors
answer
Used to detect biases or tendencies *If the sum is Positive: when the forecast on average is below the actual results Negative: when the forecasts are on average higher than the actual results
question
Qualitative Models for Forecasting
answer
Used when a situation is vague and little relevant data is available. (Based on Experience, intuition, judgement, knowledge) ex. Sales-Force Pooling Market research Customer surveys Historical analogy Delphi Method -Set up a questionarre of industry experts Management's Judgement or consensus***
question
Statistical Models for Forecasting
answer
Used when a situation is "stable" & relevant data is available (based on data and statistical techniques) Assumes that market and industry forces that influenced or generated past demand will continue to generate future demand. *if demand is purely random, statistical models cannot be able to predict what will happen, but if the demand shows some correlation to one ore more explanatory variables, a forecast can be made.
question
Role of Analytics on Forecasting
answer
Analytics is the science of examining raw data with the purpose of drawing conclusions about consumer trends and patterns. *To identify undiscovered patterns and establish hidden relationships, enabling companies to make better evidence-based business decisions and predictions. *Helps a company to be much more agile responding to customers needs and expectations
question
Business Forecast: Time Horizons
answer
Long Term (over 3 years) -New product planning, facility location& capacity, R&D -Predominantly judgmental-based forecasts Medium (1-3 yrs) -Demand forecast, aggregate planning -Mix of judgmental & statistical forecasts Short Term (Less than 1 year) -SKU Forecast, staffing levels, purchasing & inventory levels -Predominantly statistical forecasts
question
Time-Series Forecasting Models
answer
Forward Projection based on historical data or observations *Key Assumption: Historical demand patterns will continue into the foreseeable future How to: By **analyzing/identifying historical demand patterns (looking Secular trends, seasonal variations, cyclical variations, irregular behavior)
question
Naive Forecasting Model
answer
Model based on using most recent data Use the actual demand of the most recent period, to forecast the demand of the following period. *What I did last week is what I do this week is what I will do next week.
question
Moving Average Forecasting Model
answer
Model based on the arithmetic average of a specific number of historical data points. "Moves through time" -The model is updated after every new set of data becomes available and then oldest data period is dropped. * Great for short-term
question
Weighted Moving Average Model
answer
The weighted moving average model assigns weights to each data point -Decreasing weights to older data -Sum of weights must equal 1. *Assumes that the most recent points are the most important indicators of what to expect in the future.
question
Exponential Smoothing Model
answer
The Weighted Average Model between two values (the most recent demand value and its forecasted) The "weight" factor is known as the smoothing constant . Requires: Actual, Forecast, and Weight Used in Short-Term forecasting
question
Forecasting Trends
answer
When trends are present, level forecasting models will lag behind the actual values.
question
Trend-Adjusted Exponential Smoothing Model
answer
Qualitative very similar to the exponential smoothing model except it includes a component to account for the trend. Forecast = Base + Trend Concept behind smoothing with trend is to "detrend" the time series by separating the base from trend effects. *Smooth both the base and the trend
question
Linear Regression Forecasting Models
answer
A time series forecasting technique that fits a trend line to a series of data points: then, projects the results into the future.
question
Forecasting Seasonality
answer
Any regularly repeating patter represents a seasonal demand pattern Seasonality is measured by the extent to which actual values deviate from the average or mean of the data --The most frequent use to analyze seasonality is to use the seasonal indexes -At least two full season of data is required for the computation of a seasonal index
question
How to Forecast Seasonal Variation (Independent vs. In Conjunction)
answer
After the seasonal indexes are determined, the results can be used to: -Forecast demand independently or In conjuction
question
Causal Forecasting Models
answer
Identifies one or more independent variables (explanatory variables) that are related to the dependent variable being predicted Once the predictor variables are identified, a regression analysis model is used to forecast. *Causal forecasting models can be a more powerful technique than using a time-series model
question
Business Forecasts: Success Factors
answer
Success is linked to: -Continuously checking forecast accuracy -Constantly seeking ways to minimize the forecasting error -Comparing the accuracy of your forecasting model to a naive model -If you cannot beat a naive benchmark, forecasting is usually useless
question
Reasons for Forecasting failure
answer
-Not Recognizing that forecasting is integral to business planning -Not accepting that forecasted results are usually wrong -Not forecasting the right parameters -Not selecting an appropriate forecasting model -Not monitoring the accuracy of the forecast model being used -Not having enough experience & common sense -Not Synchronized through all functional ares
question
Common Mistakes in Forecasting
answer
-Forecasting what the boss wants -Not understanding relationships between variables -Misleading graphs/tables -Belief that the more complex & expensive the forecast model the more accurate it is -Good forecasts and bad forecasts will balance out -Overconfidence/Stubbornness/Wishful thinking -Not involving a broad cross section of people
question
Forecasting Management Summary
answer
-Forecasts should be used only when they can impact the decision making process -Develop contingency plans, because the actual results are always different than forecasts -Update models and revise as more information is gathered -Continuously seek for the most appropriate forecasting model -Use different approaches: Gain Confidence by validating results using different methods
Get an explanation on any task
Get unstuck with the help of our AI assistant in seconds
New