Mohlo by se vám také líbit

Informovat o naskladnění knihy

Více informací o knize Pattern Recognition with Python in 14 Days

• Parametry knihy
• Anotace
• Oblíbené z jiného soudku

Nákupem získáte 57 bodů

Anotace knihy

Fourteen days of drills on the one question every classifier answers.

This is a book of problems, not a reference. A program is shown a measurement and has to name the class it came from. Every rule is stated in as few words as it can be, shown working in a complete Python program with its real captured output, drawn as a figure, and then handed straight back as drills. Nothing is asserted that was not demonstrated: every listing was executed, and every printed output block is what the run returned.

The book prints its defeats. A bagged forest is measured losing to one tree under a fair procedure. A boosting run is measured stopping itself after seven of two hundred requested rounds. Three defensible ways of choosing the number of clusters are measured disagreeing on data with ten known classes, and none of them returns ten. Where a routine written from scratch disagrees with the library it is checked against, the disagreement is the lesson, and the chapter says which one is right and why.

What is in it

• Day 1 — Patterns, Features and Decisions. the feature vector, the decision rule, counting mistakes.


• Day 2 — Probability for Decisions. priors, likelihoods, the posterior, the rule that cannot be beaten.


• Day 3 — Costs, Risk and Refusal. loss matrices, expected risk, the reject option.


• Day 4 — The Gaussian in Many Dimensions. covariance, Mahalanobis distance, LDA and QDA.


• Day 5 — Estimating What the Rule Assumed. maximum likelihood, shrinkage, naive Bayes.


• Day 6 — Densities Without a Formula. histograms, Parzen windows, k nearest neighbours.


• Day 7 — Boundaries Fitted Directly. the perceptron, least squares, logistic regression.


• Day 8 — The Widest Corridor. margins, the hinge loss, the dual, the kernel trick.


• Day 9 — Changing the Features. principal components, whitening, Fisher's direction.


• Day 10 — Choosing Among the Features. filters, mutual information, wrappers, leakage.


• Day 11 — Measuring What Was Built. cross-validation, confusion, precision and recall, ROC, calibration.


• Day 12 — Many Classifiers, One Answer. voting, bagging, random forests, boosting, stacking.


• Day 13 — Patterns Without Labels. k-means, mixtures and EM, hierarchical and density clustering.


• Day 14 — Sequences, and a Whole System. dynamic time warping, hidden states, the pipeline.

By the numbers

• 308 worked examples, each a runnable program with its real output and a figure.


• 308 figures, drawn in grey for a black-and-white press.


• 1031 problems in five sets a chapter: review questions, worked problems, worked programming problems, and unassisted practice.


• One library built across the two weeks: prkit, a complete pattern-recognition toolkit in plain Python and NumPy, every routine checked against scikit-learn as an oracle.


• Two appendices: the whole library in one table per day, and every decision rule with its assumption, its costs and the case where it fails.

The practice problems have no answers anywhere, by design, and the preface says so.

For readers who can write Python and want to understand what a classifier is actually deciding: students, candidates preparing for examinations and interviews, and working programmers who would rather build the method than import it. No statistics beyond a first course is assumed.

Parametry knihy

• Plný název: Pattern Recognition with Python in 14 Days
• Podnázev: Master Bayes Decision Theory, Feature Design, Classifiers and Model Evaluation Through Daily Drills
• Autor: Mohammad Saqib
• Jazyk: Angličtina
• Vazba: Brožovaná
• Počet stran: 360
• EAN: 9798173925930
• ID: 53882015
• Nakladatelství: Independently published
• Hmotnost: 620 g
• Rozměry: 235 × 191 × 19 mm
• Datum vydání: 13. September 2026

---

---

---