Data Science Time Series Analysis 3

Question 1

How does SARIMA differ from a standard ARIMA model?

Accepted Answer

SARIMA adds seasonal AR, differencing, and MA terms to capture seasonal patterns

Answer

SARIMA uses machine learning instead of classical statistical methods

Answer

SARIMA is exclusively designed for monthly or quarterly data only

Answer

SARIMA does not require the time series to be stationary

Question 2

Why is standard k-fold cross-validation inappropriate for time series forecasting evaluation?

Accepted Answer

It violates temporal ordering by using future data to train models predicting the past

Answer

It is computationally too expensive for large time series datasets

Answer

Time series data is typically not normally distributed

Answer

K-fold cross-validation only works for classification problems

Question 3

What does the Granger causality test assess in time series analysis?

Accepted Answer

Whether past values of one series are useful for forecasting another series

Answer

Whether two time series have the same variance and distribution

Answer

Whether a given time series is stationary or contains a unit root

Answer

Whether significant seasonal patterns exist within a single series

Question 4

Long Short-Term Memory (LSTM) networks are well-suited for time series forecasting because:

Accepted Answer

They learn long-range dependencies and model complex non-linear patterns

Answer

They require no data preprocessing or stationarity transformations

Answer

They are always computationally faster than ARIMA-based models

Answer

They only work effectively with stationary time series data

Question 5

Cointegration between two non-stationary time series implies:

Accepted Answer

A stable long-run equilibrium relationship exists between the two series

Answer

Both series become stationary after first-order differencing individually

Answer

The two series have identical seasonal patterns and amplitudes

Answer

The series can be accurately modeled independently without any interaction

Question 6

The Facebook Prophet model for time series forecasting is based on:

Accepted Answer

An additive regression model combining trend, seasonality, and holiday effects

Answer

A state space ARIMA model with automated parameter selection

Answer

A deep neural network architecture with LSTM recurrent layers

Answer

A Kalman filter framework with Bayesian parameter estimation

Question 7

What is the 'walk-forward' (expanding window) validation strategy in time series?

Accepted Answer

Iteratively training on growing historical data and predicting the immediately following period

Answer

Using the complete dataset for both model training and testing simultaneously

Answer

Splitting the data once into a fixed training set and a fixed test set

Answer

Removing seasonal components from the data before performing model evaluation