Data Science 3

Predictive Model

A mathematical algorithm that predicts a target variable from explanatory variables.

Classification

The process of using predictor information to sort data samples into distinct classes.

Numeric Prediction

Predicting the numerical value of a dependent variable using independent variables.

Rule Induction

A process of deducing if-then rules from a data set.

Decision Tree

A model that splits data on every node leading to a leaf where the class is identified.

K-Nearest Neighbors (k-NN)

A method that classifies an unknown record based on its nearest neighbors in the training data.

Eager Learners

Models that develop a mathematical relationship between input and target variables.

Lazy Learners

Models that use a lookup table to match input variables with outcomes.

Overfitting

The tendency to tailor models too closely to training data, impacting generalization.

Pruning

The process of reducing the size of a decision tree to prevent overfitting.

Cross-Validation

A training and testing procedure that provides estimates of model generalization performance.

Confusion Matrix

A matrix that counts correct and false classifications to assess predictive capability.

Gains & Lift Charts

Charts that measure the effectiveness of a classification model against a baseline.

Test Set Validation

Holding out data to assess how well a model generalizes to unseen cases.

Generalization Performance

The ability of a model to apply to unseen data that was not part of the training set.

Domain Knowledge Validation

Sanity checking a model by getting assessments from domain experts.

Majority Vote

A method in k-NN where the class label is determined by the most common label among the nearest neighbors.

Distance Measure

A method for computing the distance between different records in k-NN.

Model Complexity

The intricacy of a model, which can lead to overfitting if not managed appropriately.

Attribute (in classification context)

Features or characteristics used to sort data samples into classes.