Statistical Methods & Forecasting Flashcards

In a time series decomposition, which component captures repeating patterns tied to calendar periods shorter than one year?

Answer: Seasonal

Seasonal components reflect regular fluctuations that repeat within a year, such as holiday sales spikes or summer demand peaks.

A market analyst runs a regression of sales (Y) on advertising spend (X) and obtains R² = 0.81. What does this mean?

Answer: 81% of sales variation is explained by advertising spend

R² measures the proportion of variance in the dependent variable explained by the independent variable(s) in the model.

Which smoothing technique assigns exponentially decreasing weights to older observations when forecasting?

Answer: Exponential smoothing

Exponential smoothing applies geometrically declining weights so recent data has greater influence on the forecast.

A p-value of 0.03 on a regression coefficient means:

Answer: We reject H₀ at the 5% significance level

A p-value below the chosen significance level (0.05) means we reject the null hypothesis that the coefficient equals zero.

The Durbin-Watson statistic is primarily used to detect:

Answer: Autocorrelation in regression residuals

The Durbin-Watson statistic tests for first-order serial correlation in the residuals of a regression model.

A forecaster uses a 3-period moving average. The last three sales values are 100, 120, and 110. What is the forecast for the next period?

Answer: 110

The 3-period moving average forecast is (100 + 120 + 110) / 3 = 110.

In hypothesis testing, a Type II error occurs when:

Answer: A false null hypothesis is not rejected

A Type II error (beta error) means failing to reject H₀ when it is actually false — a missed detection.