SCM 2160 - Midterm 2 Flashcards
What is a forecast?
A prediction of future events used for planning purposes
Based on historical data
What is a time series pattern?
Repeated observations of demand for a product/service in order of appearance
What is a horizontal time series pattern?
Fluctuation of data around a constant mean (average demand, is relatively consistent)
What is a trend time series pattern?
Systematic increase or decrease in the mean over time. Looking at a long time (not just day to day, the long run trend)
What is a seasonal time series pattern?
Repeatable pattern of increases or decrease in demand, depending on time of day, week, month, season
What is a cyclical time series pattern?
Less predictable, gradual increases in demand over long periods of time (years / decades)
What is a random time series pattern?
Un-forecastable variation in demand
What are examples of inputs to use in forecasting?
History of past demand
Notes explaining past demand
Past forecasts
Customer research
Planned promotions
Inputs from partners via CPFR
What does CPFR stand for?
Collaborative Planning, Forecasting & Replenishment
What does CPFR do?
Generate a shared forecast,
independent forecasts are generated and compared, they are adjusted until the forecasts approach consensus
What can CPFR help avoid?
It can reduce the bullwhip effect
Avoid stock outs
You can feel more certain that you can satisfy demand (retailer)
What are four popular time series methods?
Naive
Simple moving average
Weighted moving average
Exponential smoothing
What is the naive forecast method?
Only appropriate for short term forecasts
Sensitive to random variation
Ft+1 = Dt
Forecast for future period = actual demand in most recent period (t)
What is the simple moving average forecast method?
Forecast for the next period equals the average demand for n most recent periods
This smooths out random variation
How many periods do you have to wait before starting a simple moving average
Since this formula takes the average of previous periods, you need at least two previous periods to take the average of before you can start forecasting with it.
What is a weighted moving average?
The forecast for the next period equals the average demand for n most recent periods. Each observation of demand can have its own weight, with the sum = 1
The weights represent the varying amounts of influence of past demand on the forecast
How many periods do you have to wait before starting a simple moving average?
You need 3 periods of information, assigning each of those 3 different weights
What is exponential smoothing forecast method?
A weighted moving average assigning different levels of weight to recent data compared to older historical data
This requires only three data points:
- last periods forecast
- last periods demand
- smoothing parameter
What does a low ‘a’ value mean for the exponential smoothing forecast?
What does a high a value mean?
low value = more smoothing
high value = more responsive to changes
How many periods do you have to wait before starting a exponential smoothing forecast?
One period before you can start
What is forecast error?
The difference between predicted demand and actual demand
What are the four different methods to measure forecast error?
Cumulative sum of forecast errors (CFE)
Mean squared error (MSE)
Mean absolute deviation (MAD)
Mean absolute percentage deviation
What is the cumulative sum of forecast errors (CFE)
Assess total error in forecasts overtime
Calculates the error for each forecast then sum
What does it mean when CFE is really negative?
forecasts are consistently higher than demand
What does it mean when CFE is highly positive?
That the forecast is consistently lower than demand
What is the mean squared error?
Indicates on average how close forecast is to demand
Magnifies large errors