Predictive Analytics Ch. 3

You are provided 5 quarters of sales (Q1=$2,500, Q2=$2,100, Q3=$1,900, Q4=$2,000, Q5=$2,300). The moving average for those 5 quarters of sales would be:

$2,160 [(2,500 + 2,100 + 1,900 + 2,000 + 2,300) / 5 = 2,160]

What is a seed value?

(1) A method for warming up the model. (2) A selected value that allows the smoothed forecasting process to begin. (3) A method for initializing the model.

Which of the following statements are true about adaptive-response-rate single exponential smoothing (ADRES)?

(1) ADRES is a variant on simple smoothing. (2) ADRES utilizes the 'a' value to adapt to a change in the basic pattern of the data. (3) ADRES has no requirement for a smoothing constant.

The 5 basic smoothing techniques discussed in this chapter have what common characteristics?

(1) All use a form of weighted average of past observations. (2) All time-series data to be forecast are assumed to have some cycles or fluctuations that tend to recur. (3) Only the past history of the time series is necessary to produce the forecast.

Which of the following statements are true regarding simple exponential smoothing and moving averages?

(1) Both use only past values of a time series. (2) Moving averages gives equal weights to past values included in each average.

Which of the following statements are true regarding diffusion models used for new product offerings?

(1) Diffusion models can be used to identify and predict the timing of the product life cycle. (2) Diffusion models will require expert opinion to determine the correct limits on the new product's growth curves. (3) Diffusion models will have a lower and upper limit for each new product's offering.

Which of the following statements are true about diffusion models?

(1) Diffusion models have been diffused through the forecasting industry's technological innovations and new products. (2) Diffusion models can be S-curves, growth models, saturation models, or substitution curves. (3) Diffusion models can be used for new product forecasts.

Event modeling would be best described by which of the following statements?

(1) Event modeling follows the same pattern for other smoothing models. (2) Event modeling uses "events" identified as important and represented by a parameter. (3) Event modeling uses event types that are assigned its own index for a specific promotional activity.

What type of smoothing model is able to work with and adjust to data that has trend and/or seasonality?

(1) Holt's Exponential Smoothing Model (2) Winters' Exponential Smoothing Model

Which of the following statements is true about Holt's Exponential Smoothing Model?

(1) It adjusts the smoothing model for a trend in the data. (2) It is also called the Double Exponential Smoothing Holt. (3) It adds a growth factor to the smoothing equation.

Which of the following are true about Holt's Exponential Smoothing model?

(1) It can also be called linear trend smoothing. (2) It should be calculated using specialized forecasting software.

Which of the following statements are true about the limitations of the simple exponential smoothing method?

(1) Its forecasts lag behind the actual data. (2) It has no ability to adjust for any trend or seasonality in the data.

Which of the following are the most common forms of S-curves?

(1) Logistics Curve (2) Gompertz Curve (3) Bass Model

Which of the following are true about the shortcomings of moving averages?

(1) Moving averages can fail to predict peaks and troughs. (2) Moving averages can appear to identify a cycle when one is not present.

Which of the following statements are true regarding moving averages and simple exponential smoothing?

(1) Moving averages equally weight observations. (2) Exponential smoothing assumes the most recent observation will contain the most relevant information. (3) Exponential smoothing weights are made to decline exponentially with the age of the observation.Exponential smoothing

Which of the following measures are moving averages?

(1) Nine period moving average (2) Naive Model (3) Five period moving average

If a Winters' model for new car dealer sales comes up with 4 seasonal indices with values of Q1=0.95, Q2=1.04, Q3=1.05, and Q4=0.96, the following can be assumed about Q3 sales:

(1) Q3 sales are usually about 5% above an average quarter. (2) Q3 sales exhibit very strong seasonality.

Which of the following forecasting models are diffusion models?

(1) Saturation Models (2) Bass Model (3) Gompertz Curve

Which of the following are smoothing methods used in this chapter?

(1) Simple exponential smoothing (2) Adaptive-response-rate single exponential smoothing (3) Winters' exponential smoothing (4) Moving average

Which of the following statements are true about the Bass Model?

(1) The Bass Model is a relatively simple model in which only 3 parameters are chosen by the researcher. (2) The Bass Model has been used to forecast the penetration of new products in the market. (3) The Bass Model would appear most often in a graph.

Which of the following statements are true about the Gompertz Curve?

(1) The Gompertz Curve is the most used actuarial function for investigating the process of aging. (2) The Gompertz Curve is an elegant way to summarize the growth of a population. (3) The Gompertz Curve is widely used in the fields of biology and demography to model the level of populations at a given point in time.

Which of the following statements are true about the Logistics Curve model?

(1) The Logistics Curve is symmetric about its point of inflection (the upper half is a reflection of the lower half). (2) The Logistics Curve is used frequently to forecast new product sales.

Which of the following statements are true regarding the Gompertz Curve and the Logistic Curve?

(1) These models are most commonly used to forecast sales of new product and technology life cycles. (2) These models are both diffusion models. (3) These models differ in the shapes of the product curve.

Which of the following are true about simple statistical method of moving averages?

(1) This method averages the most recent values. (2) This method is most useful when the data are stationary. (3) This method can eliminate substantial randomness in the data series.

Which of the following statements about Winters' exponential smoothing model are correct?

(1) Winters' model is used for data that exhibit both trend and seasonality. (2) Winters was a student of Professor Holt and developed this modification as part of his graduate work.

If you are calculating a three month moving average, what would your interval be?

3

If you are calculating a 5 quarter moving average, what would your interval be?

5

You are provided 5 quarters of sales. If you are calculating the moving average, what would you divide the sum of the quarters of sales by?

5

What type of value is used to initialize or warm up the forecasting model?

A seed value

What type of smoothing model is able to adapt to changing circumstances for data that has little trend or seasonality?

ADRES Smoothing Model

What is adaptive-response-rate single exponential smoothing (ADRES)?

ADRES is a variant on simple smoothing that adapts to the data when there is a change in the basic pattern of the data.

The Gompertz Curve is well-known by what type of business professional?

Actuarials

When a product is new and there is no historical data, the most promising method to forecast this new product is?

Analogy

What is the third step in deseasonalizing data?

Apply a forecast method to the deseasonalized series to produce an intermediate forecast.

What is the most accurate forecasting method to use when presented with data that has a seasonal pattern?

Deseasonalizing method

T/F When working with business and economic data, it is usually a good assumption to expect no seasonality in data.

False

T/F Winters' model doesn't require the use of initial values to warm up the model because of the additional equation he added to the calculation.

False

Which one of the following models would be best for new product forecasting?

Gompertz Curve

What would be the predominant reason why you would use Holt's exponential smoothing forecast method?

It is useful for data that has a trend, but no seasonality.

What does the term exponential mean in exponential smoothing models?

It means that as you move back in time the weight assigned to the most recent observed value diminishes exponentially.

In exponential smoothing, the equation involves what type of value that is not part of the moving averages equation?

Level smoothing constant

Which of the following statements are true regarding exponential smoothing and moving averages?

Moving averages looks at values for n number of periods and exponential smoothing looks at values for all periods.

If a Winters' model for new car dealer sales, comes up with 4 seasonal indices with values of Q1=0.95, Q2=1.04, Q3=1.05, and Q4=0.96, the following can be interpreted:

Q1 sales are usually about 5% below an average quarter.

What is the final step in deseasonalizing data?

Reseasonalize the series by multiplying each deseasonalized forecast by its corresponding seasonal index.

In practice, what type of values of alpha (a) generally work best when simple exponential smoothing is the most appropriate model?

Small values

The Winters' exponential smoothing model would be best for which one of these situations?

The data is nonstationary and exhibits seasonality.

What is event modeling?

The event model adds a smoothing constant for the events identified as important in the historical data.

Which one of the following situations would lend itself well to event modeling?

The product is subject to scheduled promotions.

T/F For all exponential smoothing models, the more observations one has, the less the effect of the seed value on the final forecast.

True

What is the first step in deseasonalizing data?

Use Winters' exponential smoothing routine to create seasonal indices.

For what type of data pattern would a simple exponential smoothing model be good as a forecast method?

When data are stationary and does not have a positive or negative slope overall.

In a moving average model, every observation is given what type of weight?

equal

In the exponential smoothing forecast equation, the value 'a' is the _____.

level smoothing constant

Unlike moving average and exponential smoothing models, curve fitting models are routinely used for _____.

long-range forecasts; mid-range forecasts

Exponential smoothing gives more weight to the _____ observations and less to the _____ observations.

recent; older

In general, the weights for the exponential smoothing formula become _____ at a rate that depends on the value of (a).

smaller and smaller

Holt's Exponential Smoothing model has _____ equations and _____ smoothing constants.

three; two