ECO 309

Assume Y(1) = 232, C(1) = 1.5 and T(1)= 255, what is the value for the IRREGULAR (I) component of a for the first quarter given the following seasonal indices: 1st quarter: 0.78 = 78% 2nd quarter: 1.02 = 102% 3rd quarter: 1.12 = 112% 4th quarter: 1.08 = 108% 0.15 0.90 0.78 115 200

0.78

Time Actual (Y) Forecast (Yhat) Forecast Error 1 100 2 110 3 115 4 120 If a MA(2) model is used, what is the forecast for period 3? 5 5.2 15.2 10 10.6

10

Time Actual(Y) Forecast(Yhat) Forecast Error 1 105 2 110 3 115 4 120 ? If naïve method is used, what is the forecast for period 4? Can not be determined 110 115 125 120

115

If Y=6000 and given the trend line of 100 +50t . What is the de-trended value for t=4? (Assume a multiplicative model) 30 25 20 45 35

20

If Y(2) = 232, and trend component is T(2)=255, what is the seasonally adjusted (SA) value given the following seasonal indices: 1st quarter: 0.78 2nd quarter: 1.02 3rd quarter: 1.12 4th quarter: 1.08 227.5 Can not be determined 231.3 231 297.4

227.5

Assume Y(2) = 232,000, T(2)= 25,500, what is the de-tranded AND seasonally adjusted (SA) value ? 1st quarter: 2000 2nd quarter: -25000 3rd quarter: -13500 4th quarter: 36500 231,500 182,000 170,000 282,500 250,000

231,500

Time Data (Y) Forecast (Yhat) Error1 200 -2 210 -3 315 - 4 520 ? If a MA(3) model is used, what is the forecast error for period 4? 241.6 315 520 278.3

278.3

When checking for seasonality how many lags should I use for ACF? 12 25% of the sample size 36 Minitab will decide how many. 3 years worth of lags

3 years worth of lags

If we have the following output for monthly data, and the actual Y for February is 200, which of the following is the seasonally adjusted (SA) data? (Pick the closet answer) Period Index 1 0.31 2 0.47 3 0.88 4 1.77 5 1.91 6 1.18 7 1.02 8 1.28 9 0.93 10 0.81 11 0.60 12 0.77 425.5 166.7 645.2 Can not be determined. 203

425.5

What is the forecasted trend component for t=20 given the following trend line? T = 1700 + 150t Can not be determined without MINITAB output. 1870 4700 1700 6750

4700

Time Actual(Y) Forecast(Yhat) Forecast Error 1 100 2 110 3 115 4 120 ? If naïve method is used, what is the error for period 4? 0 15 5 10 can not be determined

5

Which of the following describe data with seasonality? A data series with only 1 significant seasonal lag A data series with 2 significant seasonal lags and the third seasonal lags with a t values greater than 1.3 A data series where the first 5-6 lags of ACF are significant A data series where the first and second lag of ACF are significant We need to check the MAPE

A data series with 2 significant seasonal lags and the third seasonal lags with a t values greater than 1.3

Which of the following is a random time series? A series whose autocorrelation function shows significant spikes at first 5-6 lags only. All of the above. A series with trend. A series with no trend but additive seasonality A series whose autocorrelation function shows no significant spikes.

A series whose autocorrelation function shows no significant spikes.

Which of the following is a trending time series? A series where is no seasonality A series whose mean is constant over time A series whose mean is changing over time A series with seasonality A series whose autocorrelation function shows only one significant spike

A series whose mean is changing over time

Which of the following is NOT correct for ACF? All of the statements are correct ACF can tell us if our data has seasonality or not. ACF can tell us if our data is random or not. ACF can tell us if the two variables are correlated or not. ACF can tell us if our data is trending or not.

ACF can tell us if the two variables are correlated or not.

How are the MAPE and MAD model selection criteria used in the model selection process? MAPE is maximized whereas MAD is minimized. Both MAPE and MAD are minimized MAPE is minimized whereas MAD is maximized. Both MAPE and MAD are maximized.

Both MAPE and MAD are minimized

If the fitted trend line is Y= 12000 - 500*t, which of the following is NOT true ? Yhat(1) is 11500. This is linear trend Can not be used for out of sample forecast. The value of Y decreases by 500 units every period. Shows constant decline in the series

Can not be used for out of sample forecast.

Which one of the following is not a component of a time series? Correlation Cylcle Seasonality Trend Irregularity

Correlation

Which of the following is NOT a smoothing technique. Correlation analysis Winter' method SIngle Double (Holt's) All of the above are smoothing techniques.

Correlation analysis

Which of the following is random? Trending data Seasonal data Trending and seasonal data Data with no predictable patterns Can't answer the question without an ACF

Data with no predictable patterns

Which of the following is NOT a quantitative forecasting method? Decomposition Smoothing MA(k) Delphi Method Naive

Delphi Method

Which method of forecasting uses all the available observations? Exponential smoothing. MAPE Moving averages. Naive.

Exponential smoothing.

Accurate forecasting can be done with inaccurate historical data. True False

False

Exponential smoothing can only be used if there is an observable trend and seasonal cycle in the data. True False

False

It is possible to forecast with zero error. True False

False

Qualitative techniques are worthless so we should only learn quantitative techniques. True False

False

Which forecasting technique assigns more weight to more recent observations? Simple average Holt's method MA(4) All methods have the same weight. Naive method

Holt's method

For what type of data do we use Single Exponential Smoothing? Trending Level Seasonal and trending Seasonal but not trending Cyclical

Level

I ran an MA(2) and a Naive model for my revenue data. My error measures for are: MA(2) MAPE: 0.58749 MAD: 1.49111 MSD: 0.31931 Naive method MAPE: 0.59295 MAD: 2.60400 MSD: 0.52015 What does this output mean? second MAD is the better method MAD is the better method MA (2) is the better method It is output and does not have an interpretation Naive is the better method

MA (2) is the better method

Which one of the following does not fall under qualitative forecasting method? Judgmental methods Delphi method Moving average methods Market research

Moving average methods

Which method of forecasting uses the most recent observation only? Naïve model squared. Naive. Exponential smoothing. Moving averages.

Naive

Which of the following is qualitative data? Inflation rate Outcome of national elections Company revenue money raised by candidates for the national election unemployment rate

Outcome of national elections

How do we want our forecast errors? MAPE less than 30%. multiplicative pattern Random Additive pattern Trending

Random

What is forecast error? MAPE, MAD and MSD It is the alpha factor in Exponential Smoothing. The forecast error has to be zero The difference between different methods. The difference between the actual (Y) and predicted (Yhat) values.

The difference between the actual (Y) and predicted (Yhat) values.

The correlation coefficient is used to determine: Which variable is important The strength of the relationship between the x and y variables A specific value of the x-variable given a specific value of the y-variable A specific value of the y-variable given a specific value of the x-variable

The strength of the relationship between the x and y variables

Below is the MINITAB output for the Autocorrelation Function on quarterly loan data. Based on these numbers, Lag ACF T LBQ 1 0.895262 4.39 21.74 2 0.788397 2.69 39.37 3 0.673311 2.68 52.85 4 0.558157 2.25 62.57 5 0.433083 2.01 68.73 6 0.402334 1.87 70.05 7. 0.379300 1.70 72.01 8. 0.341001 1.67 72.56 . . 11 -0.265639 -0.53 78.09 12 -0.351209 -0.69 84.50 13 -0.393592 -0.76 93.29 This data has seasonality and trend This data is seasonal. There is trend in this data Don't have enough data to determine trend nor seasonality. This data is random

There is trend in this data

Which of the following is true regarding measures like MSD, MAD, and MAPE? MAPE is the best one, we shouldn't use MSD and MAD. They are error measures so we want them to be 0 They complicate the forecasting process so should NOT be used. They are error measures so we want to minimize these numbers They are goodness measures so we want to maximize these numbers

They are error measures so we want to minimize these numbers

Below is the MINITAB output for the Autocorrelation Function on the number of marriages. Based on these numbers, Lag ACF T LBQ 1 0.208703 0.78 0.75 2 0.287780 1.03 2.30 3 0.226754 0.76 3.34 4 0.091159 0.29 3.53 5 0.033083 0.28 4.73 This data is cycle in this data This data has seasonality This data is random There is a trend in this data Need to have 3 years worth of data to answer this question

This data is random

Judgement is very important even for analysts with strong quantitative skills. True False

True

Which of the following methods will be more suitable to use if data is seasonal? Naive method Exponential growth Correlation MA(12) Winters' method

Winters' method

If MAPE and MAD indicate different methods, how do you choose which method to use? You should not work on this specific project anymore. You aslways trust MAD better. You always trust MAPE better. You can not choose a method You have to use your judgement.

You have to use your judgement.

A group of observations measured at successive time intervals is known as a times series additive time series model a forecast trend component

a times series

When using exponential smoothing, the smoothing constant (alpha)? indicates accuracy of previous forecast can be determined using MAD is below 0.95 and is typically greater than 0.75 in business applications. should be over one to be accurate

is below 0.95 and is typically greater than 0.75 in business applications.

The measure of forecast error which calculates the average of square of the forecast errors is known as: mean absolute error mean square error mean absolute percentage error mean error

mean square error

If there is a very strong correlation between two variables then the correlation coefficient must be any value greater than 1 much smaller than 0, if the correlation is negative much larger than 0, regardless of whether the correlation is negative or positive None of these alternatives is correct.

much smaller than 0, if the correlation is negative

Which of the following is NOT a descriptive statistic? maximum mean standard error variance p-value

p-value

Which of the following makes no sense? p-value < .10 r = .5 p-value = - .05 r= 0 r = - .95

p-value = - .05

If data for a time series analysis is collected on an annual basis ONLY, which component may be ignored? trend cyclical irregular seasonal

seasonal

Short term regular variations related to the calendar or time of day is known as? seasonality cycle randomness trend

seasonality

Simple exponential smoothing models differ from moving average models in that? They are really not different moving average models use weighted averages of the data whereas simple exponential smoothing models use simple averages. simple average uses different weights. simple exponential smoothing models use weighted averages of the data whereas moving average models use simple averages. There is no such method as moving average

simple exponential smoothing models use weighted averages of the data whereas moving average models use simple averages.