ISYE 6501 FINAL EXAM REVIEW
QUESTIONS WITH 100% CORRECT
ANSWERS
Non-parametric methods - Answer-We don't force any specific form onto the predictor
(Ex: k-nearest neighbor model)
Spline - Answer-Function of polynomials that connect to each other
Greedy Algorithm - Answer-Decisions are made step by step; at each step take one
thing that looks best and future options are not considered.
Bias - Answer-Difference between the prediction of the values by the model and the
correct values.
High Bias - Answer-Miss or minimize real effects.
Low Bias - Answer-Real effects are modeled well.
Variance - Answer-Spread of our data.
High Variance - Answer-More differentiation between predictions.
Low Variance - Answer-Less differentiation between predictions.
Overfitting - Answer-More fit to random patterns; model performs worse on a validation
or test dataset than its performance on the training dataset.
Prediction Error - Answer-Function of bias and variance.
Regularization - Answer-Prevents overfitting by penalizing model complexity, leading to
better performance on new, unseen data.
Blocking Factor - Answer-Can create variation.
Hypothesis Testing - Answer-Make decisions about a population based on sample data;
assume known underlying distribution.
Factorial Design - Answer-Determine effects of factors.
Exploration - Answer-Focusing on getting more information.
, Exploitation - Answer-Focusing on getting immediate value.
Memoryless Property - Answer-It doesn't matter what's happened in the past. All that
matters is where we are now.
Kendall Notation - Answer-The standard system used to describe and classify a
queueing node.
Balking - Answer-Unwilling to wait in line.
Entities - Answer-Things that move through simulation (e.g., bags, people, etc.).
Modules - Answer-Parts of process (e.g., queues, storage, etc.).
Outlier - Answer-Data so wrong it's noticeably different from the rest
Imputation - Answer-Estimating the missing values
Perturbation - Answer-Adding small, random changes to input data or model
parameters. Adding variability
MICE - Answer-Multivariate imputation by chained equations - can impute multiple
factor values together
Prescriptive analytics - Answer-Not only predicts what might happen but also
recommends the best course of action to achieve a desired outcome
Solution - Answer-Values for each variable
Feasible solution - Answer-Variable values that satisfy all constraints
Optimal solution - Answer-Feasible solution with the best objective value (output of an
optimization model)
Robust solution - Answer-Satisfy all solutions
Heuristic - Answer-Not guaranteed to find the best solutions, but gives very good
solutions very quickly
Parametric tests - Answer-Use exact values of data, e.g. the mean. Ex: Student's t-test.
Nonparametric tests - Answer-Use ranks of data, e.g. the median. Use even when
nothing is known about underlying distribution
QUESTIONS WITH 100% CORRECT
ANSWERS
Non-parametric methods - Answer-We don't force any specific form onto the predictor
(Ex: k-nearest neighbor model)
Spline - Answer-Function of polynomials that connect to each other
Greedy Algorithm - Answer-Decisions are made step by step; at each step take one
thing that looks best and future options are not considered.
Bias - Answer-Difference between the prediction of the values by the model and the
correct values.
High Bias - Answer-Miss or minimize real effects.
Low Bias - Answer-Real effects are modeled well.
Variance - Answer-Spread of our data.
High Variance - Answer-More differentiation between predictions.
Low Variance - Answer-Less differentiation between predictions.
Overfitting - Answer-More fit to random patterns; model performs worse on a validation
or test dataset than its performance on the training dataset.
Prediction Error - Answer-Function of bias and variance.
Regularization - Answer-Prevents overfitting by penalizing model complexity, leading to
better performance on new, unseen data.
Blocking Factor - Answer-Can create variation.
Hypothesis Testing - Answer-Make decisions about a population based on sample data;
assume known underlying distribution.
Factorial Design - Answer-Determine effects of factors.
Exploration - Answer-Focusing on getting more information.
, Exploitation - Answer-Focusing on getting immediate value.
Memoryless Property - Answer-It doesn't matter what's happened in the past. All that
matters is where we are now.
Kendall Notation - Answer-The standard system used to describe and classify a
queueing node.
Balking - Answer-Unwilling to wait in line.
Entities - Answer-Things that move through simulation (e.g., bags, people, etc.).
Modules - Answer-Parts of process (e.g., queues, storage, etc.).
Outlier - Answer-Data so wrong it's noticeably different from the rest
Imputation - Answer-Estimating the missing values
Perturbation - Answer-Adding small, random changes to input data or model
parameters. Adding variability
MICE - Answer-Multivariate imputation by chained equations - can impute multiple
factor values together
Prescriptive analytics - Answer-Not only predicts what might happen but also
recommends the best course of action to achieve a desired outcome
Solution - Answer-Values for each variable
Feasible solution - Answer-Variable values that satisfy all constraints
Optimal solution - Answer-Feasible solution with the best objective value (output of an
optimization model)
Robust solution - Answer-Satisfy all solutions
Heuristic - Answer-Not guaranteed to find the best solutions, but gives very good
solutions very quickly
Parametric tests - Answer-Use exact values of data, e.g. the mean. Ex: Student's t-test.
Nonparametric tests - Answer-Use ranks of data, e.g. the median. Use even when
nothing is known about underlying distribution
