Linear Stationary Processes and White Noise
1. What is a purely random process also known as?
A. Brownian motion
B. Autocorrelation
C. White noise
D. Random walk
Answer: C
2. In a purely random process, the mean is:
A. Non-zero
B. Increasing over time
C. Zero
D. Undefined
Answer: C
3. A purely random process assumes which distribution?
A. Poisson
B. Normal
C. Binomial
D. Exponential
Answer: B
4. A linear stationary process can be expressed as:
A. A deterministic sum
B. A weighted sum of past white noises
C. A sine wave
D. A polynomial equation
Answer: B
5. For stationarity in a linear process, weights must:
A. Be random
B. Decrease rapidly
C. Be constant
D. Follow a normal distribution
Answer: B
Moving Average (MA) Models
6. Moving average (MA) processes are frequently used in:
A. Chemistry
B. Biology
C. Econometrics
D. Geography
Answer: C
7. MA(q) stands for:
A. Multi-approach model
B. Moving average of order q
C. Maximum average
D. None
Answer: B
8. The white noise term in MA(q) is scaled so that:
A. β₀ = 2
B. β₀ = 0
C. β₀ = 1
D. β₀ = -1
Answer: C
9. The autocorrelation function of MA(q) becomes zero when:
A. lag < q
B. lag > q
C. lag = q
D. q = 0
Answer: B
10. The invertibility condition for MA(1) is:
A. |β₁| > 1
B. |β₁| < 1
C. β₁ = 0
D. |β₁| = 1
Answer: B
Weighted Moving Average (WMA)
11. In WMA, recent data points are given:
A. Less weight
B. Equal weight
C. No weight
D. Greater weight
Answer: D
12. In simple moving average, weights are:
A. Increasing
B. Random
C. Equal
D. Zero
Answer: C
13. WMA is most useful when:
A. All data is outdated
B. Future data is known
C. Recent changes are more relevant
D. No variance exists
Answer: C
Autoregressive (AR) Models
14. AR(p) process depends on:
A. Random noise only
B. Past p observations
C. Future values
D. Trends only
Answer: B
15. AR(1) model has how many lag terms?
A. 1
B. 0
C. 2
D. Infinite
Answer: A
16. For AR(p) stationarity, the roots of the characteristic equation must be:
A. Inside unit circle
B. On unit circle
C. Outside unit circle
D. Complex
Answer: C
17. AR(2) process includes:
A. 1 lag term
B. 2 lag terms
C. White noise only
D. No memory
Answer: B
18. Large negative α₁ in AR(1) leads to:
A. Smooth trend
B. Oscillatory zig-zag
C. Constant series
D. No effect
Answer: B
ARMA Models
19. ARMA combines:
A. Only AR
B. Only MA
C. AR and MA
D. Trend and seasonality
Answer: C
20. ARMA is preferred due to:
A. Complexity
B. Parsimony
C. Inaccuracy
D. Irregularity
Answer: B
21. ARMA(p, q) has:
A. p moving average terms
B. q autoregressive terms
C. p AR and q MA terms
D. None of the above
Answer: C
22. ARMA models are stationary and invertible when:
A. All roots lie inside unit circle
B. Roots lie outside unit circle
C. No roots exist
D. Roots are imaginary
Answer: B
ARIMA Models
23. What does “I” in ARIMA stand for?
A. Integration
B. Independence
C. Invertibility
D. Intensity
Answer: A
24. Differencing helps to:
A. Make data non-linear
B. Remove seasonality
C. Stabilize variance
D. Make series stationary
Answer: D
25. ARIMA(p,d,q) becomes ARMA if:
A. d = 1
B. d = 2
C. d = 0
D. d > 0
Answer: C
26. ARIMA(0,1,0) represents:
A. Pure noise
B. Random walk
C. Seasonal trend
D. Invertible model
Answer: B
27. To achieve stationarity in ARIMA:
A. Add noise
B. Take moving average
C. Apply differencing
D. Increase variance
Answer: C
SARIMA Models
28. SARIMA adds which component to ARIMA?
A. Autocovariance
B. Spectral density
C. Seasonality
D. Residuals
Answer: C
29. Seasonal differencing involves:
A. Removing white noise
B. Differencing at seasonal lags
C. Integration of ARMA
D. Removing trend
Answer: B
30. In SARIMA(p,d,q)(P,D,Q)s, “s” stands for:
A. Stationarity
B. Season
C. Shift
D. Standard deviation
Answer: B
31. SARIMA models are ideal for:
A. Non-repetitive trends
B. Non-linear patterns
C. Data with seasonal cycles
D. Gaussian noise
Answer: C
32. SARIMA handles both:
A. Trend and noise
B. Linear and non-linear components
C. Seasonal and non-seasonal behavior
D. MA and AR terms only
Answer: C
Applications & Behavior
33. Autocorrelations in MA(q) drop to zero after:
A. Lag 1
B. Lag q
C. Lag < q
D. Never
Answer: B
34. MA models allow:
A. Dependence on future values
B. Infinite lag
C. Past shock effects
D. Trend prediction
Answer: C
35. AR models allow:
A. Forecasting with no data
B. Only white noise modeling
C. Regressing on previous values
D. Perfect predictions
Answer: C
36. If α₁ = 0.9 in AR(1), the series is:
A. Very volatile
B. Zig-zag
C. Smooth with slow change
D. Random
Answer: C
37. If β₁ = 0.8 in MA(1), the invertibility condition:
A. Is violated
B. Is satisfied
C. Does not apply
D. Is unknown
Answer: B
38. In AR(2), ρ₁ and ρ₂ are calculated from:
A. White noise
B. Variance
C. Coefficients α₁ and α₂
D. Trends
Answer: C
Conceptual
39. Parsimony in modeling means:
A. Overfitting
B. Using minimal parameters
C. Ignoring outliers
D. Removing trends
Answer: B
40. The backward shift operator is used in:
A. Regression
B. ARIMA models
C. MA models
D. WMA
Answer: B
41. ARIMA assumes:
A. No structure
B. Nonlinear dependence
C. Linearity
D. Seasonality
Answer: C
42. A non-invertible process can’t:
A. Predict future
B. Be stationary
C. Be graphed
D. Use residuals
Answer: A
43. Random walk is a special case of:
A. ARIMA
B. MA
C. WMA
D. ARMA
Answer: A
44. Which model best handles monthly sales data with trend and seasonality?
A. ARIMA
B. SARIMA
C. MA
D. AR
Answer: B
45. One difference applied to ARIMA means:
A. d = 1
B. q = 1
C. p = 1
D. Differencing fails
Answer: A
46. ARIMA model is best for:
A. Short-term seasonal forecast
B. Long-term periodic forecast
C. Short-term stationary series
D. Forecasting exact values
Answer: C
Miscellaneous / Higher Order Thinking
47. What happens if roots lie inside unit circle in AR process?
A. Model is invertible
B. Model is stationary
C. Model is non-stationary
D. All are true
Answer: C
48. In MA(2), how many autocorrelations are non-zero?
A. Infinite
B. 0
C. 2
D. Depends on α
Answer: C
49. In AR(1), which of the following affects the smoothness of the series?
A. Mean
B. Variance
C. α₁
D. β₀
Answer: C
50. Which of these statements is true?
A. WMA uses only equal weights
B. ARIMA can handle non-seasonal data only
C. SARIMA is a generalized ARIMA
D. AR(1) = MA(1)
Answer: C
