AN 420G FINAL EXAM QUESTIONS ANSWERED CORRECTLY LATEST UPDATE 2026

AN 420G FINAL EXAM QUESTIONS ANSWERED CORRECTLY LATEST UPDATE 2026 Neural Network - Answers A model that combines inputs in a complex flexible structure to make predictions Input Layer - Answers The first layer that takes in raw data where input equals output Hidden Layer - Answers Intermediate layers that transform inputs using weighted sums and activation functions Output Layer - Answers Final layer that produces prediction or classification Node - Answers A processing unit in a neural network that receives inputs and produces output Weights - Answers Parameters that determine the importance of inputs and are adjusted during training Bias - Answers Additional parameter added to nodes to shift the activation function Activation Function - Answers Function applied to weighted sum (e.g., logistic) to produce node output Logistic Function - Answers A function that maps values between 0 and 1 for classification Training - Answers Process of adjusting weights to minimize prediction error Backpropagation - Answers Process of propagating error backward to update weights Learning Rate - Answers Parameter controlling how much weights change each iteration Overfitting - Answers When model learns noise instead of pattern and performs poorly on new data Validation Data - Answers Data used to evaluate model performance during training Scaling - Answers Transforming variables (often 0-1) to improve model performance Dummy Variables - Answers Binary variables created for categorical data Iteration - Answers One full pass through training data updating weights MLPClassifier - Answers Scikit-learn tool for building neural networks Deep Learning - Answers Neural networks with many layers used for complex tasks like image recognition Convolutional Neural Network (CNN) - Answers Neural network that processes grouped inputs like image pixels Discriminant Analysis - Answers Statistical classification method based on distance to class centers Centroid - Answers Mean vector of predictors for a class Euclidean Distance - Answers Straight-line distance between two points Mahalanobis Distance - Answers Distance that accounts for correlation between variables Classification Function - Answers Equation used to assign observations to classes Decision Function - Answers Difference between classification scores in binary classification Linear Discriminant Analysis (LDA) - Answers A linear method for separating classes Prior Probabilities - Answers Probability of belonging to each class before observing data Confusion Matrix - Answers Table showing correct and incorrect classifications Propensity Score - Answers Probability that a record belongs to a class Multivariate Normality - Answers Assumption that predictors follow a normal distribution Outliers - Answers Extreme values that can distort results Misclassification Cost - Answers Cost associated with incorrect classification Ensemble Method - Answers Combining multiple models to improve prediction accuracy Bagging - Answers Method using bootstrap samples and averaging predictions Bootstrap Sampling - Answers Sampling with replacement from dataset Boosting - Answers Method focusing on correcting errors from previous models Random Forest - Answers Ensemble of decision trees using bagging AdaBoost - Answers Boosting algorithm that adjusts weights of observations Majority Voting - Answers Choosing most common prediction in classification Averaging - Answers Taking mean prediction for regression Variance Reduction - Answers Decreasing variability in predictions using ensembles Wisdom of Crowds - Answers Idea that combined predictions outperform individuals Overfitting Reduction - Answers Benefit of ensembles in improving generalization Association Rules - Answers Patterns showing relationships between items Market Basket Analysis - Answers Analysis of items purchased together Antecedent - Answers "If" part of rule Consequent - Answers "Then" part of rule Support - Answers Percentage of transactions containing an itemset