ISYE 6501 Models Questions with 100%
Verified Correct Answers
I have a time series and want to forecast future values using past values and past
forecast errors. What model should I consider?
ARIMA
Common uses
Sales forecasting
Demand forecasting
Energy consumption
Look for
Historical time series
Forecasting future values
Not for
Classification
Clustering
Tabular prediction without a time component
I want to forecast a time series and believe recent observations should matter more than
older observations.
Exponential Smoothing
Common uses
Sales forecasts
Inventory planning
Call center demand
Look for
, Time series
Recent data is most important
I want to predict future volatility (uncertainty) rather than future values.
GARCH
Common uses
Stock volatility
Risk management
Financial returns
Look for
Volatility
Risk
Variance changes over time
My outcome is Yes/No and I need probabilities and interpretable coefficients.
Logistic Regression
Common uses
Churn prediction
Loan default
Disease diagnosis
Fraud detection
Look for
Binary outcome
Need interpretability
Need probabilities
My outcome is numeric and I believe the relationship is approximately linear.
Verified Correct Answers
I have a time series and want to forecast future values using past values and past
forecast errors. What model should I consider?
ARIMA
Common uses
Sales forecasting
Demand forecasting
Energy consumption
Look for
Historical time series
Forecasting future values
Not for
Classification
Clustering
Tabular prediction without a time component
I want to forecast a time series and believe recent observations should matter more than
older observations.
Exponential Smoothing
Common uses
Sales forecasts
Inventory planning
Call center demand
Look for
, Time series
Recent data is most important
I want to predict future volatility (uncertainty) rather than future values.
GARCH
Common uses
Stock volatility
Risk management
Financial returns
Look for
Volatility
Risk
Variance changes over time
My outcome is Yes/No and I need probabilities and interpretable coefficients.
Logistic Regression
Common uses
Churn prediction
Loan default
Disease diagnosis
Fraud detection
Look for
Binary outcome
Need interpretability
Need probabilities
My outcome is numeric and I believe the relationship is approximately linear.