HomeBrowse About Blog

Plan your day, study, read, write and track your habits in one app.

Follow MugUp

Product

Features
Compare
Free Tools
Deck Library
MugUp for Discord

Community

Blog
Changelog
Feature Requests
Deck Requests

Company

Home
About
Contact

© 2026 MugUp. All rights reserved.

Terms of Service Privacy Policy Cookie Policy

Browse decks
Pattern Recognition (COMP 5013)

Public deck

Pattern Recognition (COMP 5013)

COMP 5013 at Lakehead, built from the course textbook (An Introduction to Statistical Learning with Python): regression, classification, resampling, regularization, splines and GAMs, trees, SVMs, neural networks, PCA and clustering, plus the Python used in the labs and the mixture and ICA material from the course outline.

School

RShared by Rushan

0 Play Sign in to clone

You’re reading a free preview. Sign up to see every topic and question — and to play this deck as games.

See the whole deck

Suggested order

Learn from the top down. Topics side by side can be studied in any order. Cloning the deck brings this order with it.

Topics

Open a topic to read through its content, then play just that topic.

What Statistical Learning Is

ISLP section 2.1: the model Y = f(X) + ε, why we estimate f (prediction or inference), parametric and non-parametric methods, the flexibility and interpretability trade-off, and the main kinds of learning problem.

35 total questions

Model Accuracy and the Bias-Variance Trade-Off

ISLP section 2.2: measuring fit with MSE and error rate, training versus test error, overfitting, the bias-variance decomposition, the Bayes classifier and the K-nearest neighbors classifier.

43 total questions

Simple and Multiple Linear Regression

Least squares, the accuracy of the coefficient estimates (standard errors, confidence intervals, t-tests, p-values), RSE and R², and multiple regression with the F-test, variable selection and prediction intervals (ISLP 3.1-3.2).

Locked · 48 questions inside

Linear Regression Extensions and Problems

Qualitative predictors and dummy variables, interaction terms and polynomial regression, the six potential problems with a linear fit and how to spot them, the Advertising marketing plan, and linear regression versus KNN regression (ISLP 3.3-3.5).

Locked · 48 questions inside

Logistic Regression

Why linear regression is a poor fit for classification, and how logistic regression models a probability through the log odds. Covers maximum likelihood, reading coefficient tables, making predictions, confounding in the Default data and multinomial logistic regression (ISLP 4.1-4.3).

Locked · 46 questions inside

LDA, QDA and Naive Bayes

Generative classifiers that model each class's predictor distribution and turn it into a class probability with Bayes' theorem. Covers LDA for one and many predictors, confusion matrices, sensitivity and specificity, thresholds and the ROC curve, QDA and naive Bayes (ISLP 4.4).

Locked · 44 questions inside

Comparing Classifiers and Poisson Regression

How logistic regression, LDA, QDA, naive Bayes and KNN relate mathematically and which one wins in which kind of data. Then Poisson regression for count responses, using the Bikeshare data, and generalized linear models in general (ISLP 4.5-4.6).

Locked · 45 questions inside

Cross-Validation and the Bootstrap

Estimating test error with the validation set approach, LOOCV and k-fold cross-validation, and measuring the uncertainty of an estimate with the bootstrap. Covers ISLP sections 5.1 and 5.2.

Locked · 44 questions inside

Subset Selection, Cp, AIC and BIC

Choosing a subset of predictors with best subset, forward stepwise and backward stepwise selection, then picking the model size with Cp, AIC, BIC, adjusted R², a validation set or cross-validation. Covers ISLP section 6.1.

Locked · 41 questions inside

Ridge Regression and the Lasso

The two shrinkage methods from ISLP section 6.2: their penalties, how λ trades bias for variance, why the lasso gives exact zeros and ridge does not, which one to expect to win, and how λ is chosen.

Locked · 45 questions inside

PCR, PLS and High-Dimensional Data

Dimension reduction for regression from ISLP sections 6.3 and 6.4: principal components regression, partial least squares, and what goes wrong (and what to report) when p is close to or larger than n.

Locked · 44 questions inside

Polynomials, Step Functions and Regression Splines

Moving beyond a straight line with polynomial regression, step functions and basis functions, then piecewise polynomials, cubic and natural splines, and how to choose the knots (ISLP 7.1-7.4).

Locked · 44 questions inside

Smoothing Splines, Local Regression and GAMs

Smoothing splines and how λ and the effective degrees of freedom control them, local regression and its span, and generalized additive models for regression and classification (ISLP 7.5-7.7).

Locked · 44 questions inside

Regression and classification trees from ISLP section 8.1: how recursive binary splitting grows a tree, how cost complexity pruning and cross-validation cut it back, the Gini index and entropy, and how trees compare with linear models.

Locked · 45 questions inside

Bagging, Random Forests, Boosting and BART

The four tree ensembles in ISLP section 8.2: bagging and out-of-bag error, random forests, boosting and its three tuning parameters, and Bayesian additive regression trees, plus how the four differ.

Locked · 43 questions inside

Maximal Margin and Support Vector Classifiers

Hyperplanes, separating hyperplanes, the maximal margin classifier and its support vectors, and the support vector classifier with slack variables and ISLP's budget C (ISLP sections 9.1-9.2).

Locked · 42 questions inside

Kernels and Support Vector Machines

Enlarging the feature space, the inner-product form of the classifier, linear, polynomial and radial kernels, the Heart data ROC comparison, SVMs for more than two classes, and the hinge-loss link to logistic regression (ISLP sections 9.3-9.5).

Locked · 41 questions inside

Neural Networks: Structure and Fitting

Single layer and multilayer neural networks (ISLP 10.1, 10.2), when deep learning is worth using (10.6), how networks are fit with gradient descent, backpropagation and regularization (10.7), and interpolation and double descent (10.8).

Locked · 47 questions inside

CNNs, RNNs and Document Classification

Convolutional neural networks for images (ISLP 10.3), bag-of-words document classification on the IMDb reviews (10.4), and recurrent neural networks for documents and the NYSE time series (10.5).

Locked · 47 questions inside

Principal Components Analysis and Matrix Completion

Why unsupervised learning is hard to evaluate, how PCA finds loadings and scores, how to read a biplot and a scree plot, when to scale, and how PCA fills in missing values (matrix completion and recommender systems).

Locked · 49 questions inside

K-Means and Hierarchical Clustering

How K-means minimizes within-cluster variation and why it needs several random starts, how to read and cut a dendrogram, the four linkages, choosing a dissimilarity measure and scaling, and the practical cautions ISLP gives about trusting clusters.

Locked · 46 questions inside

Python Tools for the Labs

The Python used in the course's Colab assignments: numpy arrays, pandas data frames, statsmodels fits, the ISLP helpers, and the sklearn estimators for classification, resampling, regularisation, trees, SVMs, PCA and clustering.

Locked · 49 questions inside

Gaussian Mixtures, EM and Independent Component Analysis

Mixture of Gaussians, the EM algorithm, and independent component analysis, compared with K-means and PCA. These come from the course outline rather than the ISLP textbook.

Locked · 47 questions inside

21 more topics in this deck

Create a free account to open every topic — and play the whole deck without copying it.

Sign up free Log in