Week 2 / Chapter 2: Forecasting Flashcards
Forecast
A statement about the future
Business forecasting
Demand, profits, costs, prices, interest rates, stock prices, general economic or political indicators, etc.,
In Operations management
Demand forecast is a major input to many decisions
Many management decisions are “________” decisions for the future
planning
Forecasts anticipate _______
Forecasts reduce ______
future uncertainty (less uncertainty = better decision)
Therefore, forecasts are the basis of corporate planning:
Long-term: products/services, processes, capacity, facilities layout, equipment, location, etc.
Short/intermediate term: operations planning, purchasing, inventory levels, workforce levels, scheduling, production, etc
Planning decisions are made _______
Forecasts are needed __________
continuously
continuously
Main features of forecasting
Assumes causal system that existed in the past and will do so in the future
Forecasts are rarely “perfect” because of randomness
Forecasts are more accurate for groups vs. individual items
Forecast accuracy decreases as time horizon increases
Good forecasting has 3 components, these are?
1) Accurate
2) Affordable
3) Timely
6 Steps to creating a forecast
1) Determine the purpose of the forecast
2) Establish a time horizon
3) Gather and analyze data
4) Select a forecasting technique
5) Prepare the forecast
6) Monitor the forecast
Subjective Forecasting Techniques:
Rely solely on judgments and opinions of experts
Required when data is not available
Objective Forecasting Techniques:
Time Series Models:
Associative/Causal models:
Time Series Models:
use historical data to predict the future
Associative/Causal models:
use explanatory variables to predict the future
Subjective Forecasts are applicable when?
Applicable when:
- There is no time to gather data
- Data is obsolete (e.g. due to economic changes)
- No data available (e.g. new products)
Judgmental Methods:
Executive opinions Sales force composite Consumer surveys Outside opinion Opinions of managers and staff: Panel Consensus, Delphi Technique
Time Series Analysis (definition)
A chronological series of observations taken over time
Details of Time Series Analysis
Assumes that future values can be calculated only from past data
In demand forecasting, past data on demand is used to predict future demand
Let At = Actual demand in period t
Ft = Forecast for period t
Basic Question: Given A1, A2, A3,… At-1, What is Ft?
Components of a time series:
Identify the behaviour of past data, look for the patterns!
1) Level
2) Trend
3) Seasonality
4) Cyclical
5) Random Variations
1) Level
short-term movement around a fixed average
2) Trend
long-term movement in data (up/down)
3) Seasonality
short-term regular variations in data (less than a year)
4) Cyclical
long-term regular variations in data (more than a year)
5) Random Variations
caused by chance / unusual circumstances
Techniques for Averaging
Works best if the series is fluctuating around some average (stationary)
Moving Averages
Exponential Smoothing
Moving Average Equation
This is the classic way to calculate an average. Add up all the numbers and divide by the amount of numbers
Ft = Demand in pervious n periods / n
10 + 5 + 6 + 11 + 9 / 5 = 6.2 average
Most times it asks for the three-period moving average in which you would take the 3 most recent data points (the higher periods) and average those
Weighted Moving Average Equation (way to calculate GPA)
Ft = (weight of period n) x (demand of period n) / Weights
AKA Grade x Credits / Weight
Still legs behind, but not as bad as moving average - it is sharper than moving average
Exponential Smoothing Equation
Ft= Ft-1 + alpha(At-1 - Ft-1)
F = Forecast A = Actual Alpha = Smoothing Constant
Moving Average theory
Average of last few actual data values, updated each period:
- Easy to calculate and understand
- Lags behind changes
Moving average
Choose number of periods to include:
- Fewer data points = more sensitive to changes
- More data points = smoother, less responsive
Moving average Exmaple
Literally calculates like a normal avergae
F3 = (43+40+41) / 3 = 41.33
Weighted Moving Average - Example
Compute a 4-period weighted moving average forecast for period 6 using a weight of 0.4 for the most recent period, 0.3 for the next, 0.2 for the next, and 0.1 for the next.
Period 1: 42 Period 2: 40 Period 3: 43 Period 4: 40 Period 5: 41
F6 = 0.40(41) + 0.30(40) + 0.20(43) + 0.10(40) = 41.0
When using the Weighted Moving Average the weights must add up to ____%
100%
If it is above 100% then an increase in terms of forecasting of observation of actual data
If it is less than 100% then the odds are decreasing
Exponential Smoothing
- Sophisticated weighted moving average
- New forecast is based on the previous forecast plus a percentage of the difference between that forecast and the previous actual value
- Subjectively choose smoothing constant alpha
- Alpha ranges from 0 to 1 (commonly .05 to .5)
- A forecast of zero means that it would be the same as forecast from the last period
- Widely used
- Easy to use
- Easy to alter weighting
Exponential Smoothing Forumla
Ft = Ft-1 + Alpha(At-1 - Ft-1)
Where F = Forecast
A = Actual
Alpha = % of the error applied to the previous
forecast to generate a new forecast. Between
0-1
How to calculate the Error Term in exponential smoothing
Actual - Forecast
The larger the smoothing constant (alpha) the more ____________________
responsive the forecast!
Choosing Alpha
When demand is fairly stable, use a lower value for Alpha
-Smooths out random fluctuations
When demand increasing or decreasing, use a higher value for Alpha
-More responsive to real changes
Try to find balance:
- Trial and error
- Can change over time
A moving average forecast tends to be more responsive to changes in the data series when more data points are included in the average.
False
As compared to a simple moving average, the weighted moving average is more reflective of the recent changes.
True
A smoothing constant of .1 will cause an exponential smoothing forecast to react more quickly to a sudden change than a value of .3 will
False
Average techniques
Easy to understand and interpret
Smooth the variations in actual data points
Lags behind trends
