Statistics 2

Front 1

Naive Model

Back 1

Assumes that the value of the series next period will be the same as it is this period.

Front 2

Warm-up sample

Back 2

First half of the data, used to fit the forecasting model.

Front 3

Warmed up

Back 3

Running the model through the first part.

Front 4

Forecasting sample

Back 4

The second half of the data used to test the model.

Front 5

Accuracy

Back 5

Not critical in the warm-up sample, but in the forecasting sample, since the pattern of the data often changes over time.

Front 6

Forecasting sample

Back 6

Used to evaluate how well the model tracks which changes.

Front 7

Warm-up sample

Forecasting Sample

Back 7

Accuracy is not ... , But in the ...

Front 8

Moving Average

Back 8

An indicator frequently used in technical analysis showing average value of a quantity over a set period.

Front 9

Moving averages

Back 9

Generally used to measure momentum and define areas of possible support and resistance

Front 10

Fluctuations or noise

Interpretation

Back 10

Moving averages are used to emphasize the direction of a trend and to smooth out ... That can confuse ...

Front 11

Upward momentum

Back 11

Confirmed when a short term average crosses above longer-term average.

Front 12

Downward momentum

Back 12

Confirmed when a short-term average crosses below a long-term average.

Front 13

Simple moving average

Back 13

The data from the consecutive time periods being considered are given equal weights in determining forecast value

Front 14

Forecast value

Back 14

Mean of the last N data points.

Front 15

Experimentation

Accuracy

Back 15

The only way to decide on the number of periods is by ... One determine which one has a better ...

Front 16

Weighted Moving average

Back 16

Subscribes to the principle in forecasting that recent data contain more information than older data

Front 17

Simple exponential smoothing

Back 17

Works like ab automatic pilot or thermostat

Front 18

Increased

Back 18

If forecast errors are positive, the forecast are ...

Front 19

Reduced

Back 19

If forecast errors are negative, the forecast are ...

Front 20

Zero

Back 20

Process of adjustment continues unless the errors reach ...

Front 21

Old forecast plus a fraction of the error

Back 21

The new forecast is equal to

Front 22

Exponential smoothing parameter

Back 22

Fraction of the error

Front 23

Data storage requirements are minimal

Back 23

Advantage over the moving average

Front 24

Slow

Back 24

Important limitation: assumes that any change in the mean of the tine series will be ...

Front 25

Best fitting

Back 25

One that gives minimum MSE

Front 26

Amount of noise or randomness in series

Back 26

The greater the noise, the smaller the particular a should be to avoid overreaction to purely random fluctuations.

Front 27

Stability of the mean

Back 27

Mean is relatively constant, a should be small

Front 28

Large

Back 28

If mean is changing, a should be ... to keep up with the changes.

Front 29

Robust model

Back 29

Gives good performance on many different kinds of time series, especially those that conyain a great deal of noise.

Front 30

Constant level

Back 30

Assumes no trend at all

Front 31

Cinstant level

Back 31

The time series us assumed to have a relativity constant mean

Front 32

Constant

Back 32

The forecast for any period in the future us a horizontal line.

Front 33

Moving Average and Exponential Smoothing

Back 33

Forecast methods of constant level

Front 34

Linear Trending

Back 34

Forecasts a straight line trend for any period in the future.

Front 35

Time series regression and Smoothing linear trend

Back 35

Forecast methods of linear trend

Front 36

Exponential Trend

Back 36

Forecasts that the amount of growth will increase continuously.

Front 37

Exponential

Back 37

At long horizons, these trends become unrealistic.

Front 38

Damped Trend

Back 38

Developed for longer-range forecasting

Front 39

Damped Trend

Back 39

The amount of trend extrapolated declines each period in a ...

Front 40

Damped Trend

Back 40

Eventually the trend dies out and the forecasts become a horizontal line.

Front 41

Seasonal Patterns

Back 41

May exist in the data

Front 42

Additive Seasonal pattern

Multiplicative Seasonal Pattern

Back 42

2 seasonal patterns

Front 43

Additive Seasonal Pattern

Back 43

Assumes that the seasonal fluctuations are of constant size

Front 44

Additive Seasonal Pattern

Back 44

Seasonal pattern less common in business data.

Front 45

Multiplicative Seasonal Pattern

Back 45

Assumes that the seasonal fluctuations are proportional to the data.

Front 46

Multiplicative Seasonal Pattern

Back 46

As the trend increases the seasonal fluctuations get larger.

Front 47

Causal Methods

Back 47

Attempt to find a relationship between the variable to be forecast and one or more other variables.

Front 48

Sales might be forecast as a function of advertising and price

Back 48

Example of causal method

Front 49

Sales

Back 49

Dependent on in any factors

Front 50

(1) common to be ranked differently depending on accuracy measure used

(2) up to the manager to decide which accuracy measure is most appropriate for his or her application.

(3) evaluating forecast accuracy is needed since large errors cab be extremely disruptive

Back 50

Three evaluations for forecast accuracy

Front 51

Mean Absolute Forecast Error/Mean Absolute Deviation

Back 51

Gives equal weight to each error.

Front 52

Mean Absolute Deviation

Back 52

MAD

Front 53

Mean Absolute Percentage Error

Back 53

MAPE

Front 54

Mean Square Error

Back 54

MSE

Front 55

Mean Square Error

Back 55

Gives more weight to large errors.