Python机器学习入门

Preface1. IntroductionWhy Machine Learning?Problems Machine Learning Can SolveKnowing Your Task and Knowing Your DataWhy Python?scikit-learnInstalling scikit-learnEssential Libraries and ToolsJupyter NotebookNumPySciPymatplotlibpandasmglearnPython 2 Versus Python 3Versions Used in this BookA First Application: Classifying Iris SpeciesMeet the DataMeasuring Success: Training and Testing DataFirst Things First: Look at Your DataBuilding Your First Model: k-Nearest NeighborsMaking PredictionsEvaluating the ModelSummary and Outlook2. Supervised LearningClassification and RegressionGeneralization, Overfitting, and UnderfittingRelation of Model Complexity to Dataset SizeSupervised Machine Learning AlgorithmsSome Sample Datasetsk-Nearest NeighborsLinear ModelsNaive Bayes ClassifiersDecision TreesEnsembles of Decision TreesKernelized Support Vector MachinesNeural Networks (Deep Learning)Uncertainty Estimates from ClassifiersThe Decision FunctionPredicting ProbabilitiesUncertainty in Multiclass ClassificationSummary and Outlook3. Unsupervised Learning and PreprocessingTypes of Unsupervised LearningChallenges in Unsupervised LearningPreprocessing and ScalingDifferent Kinds of PreprocessingApplying Data TransformationsScaling Training and Test Data the Same WayThe Effect of Preprocessing on Supervised LearningDimensionality Reduction, Feature Extraction, and Manifold LearningPrincipal Component Analysis (PCA)Non-Negative Matrix Factorization (NMF)Manifold Learning with t-SNEClusteringk-Means ClusteringAgglomerative ClusteringDBSCANComparing and Evaluating Clustering AlgorithmsSummary of Clustering MethodsSummary and Outlook4. Representing Data and Engineering FeaturesCategorical VariablesOne-Hot-Encoding (Dummy Variables)Numbers Can Encode CategoricalsBinning, Discretization, Linear Models, and TreesInteractions and PolynomialsUnivariate Nonlinear TransformationsAutomatic Feature SelectionUnivariate StatisticsModel-Based Feature SelectionIterative Feature SelectionUtilizing Expert KnowledgeSummary and Outlook5. Model Evaluation and ImprovementCross-ValidationCross-Validation in scikit-learnBenefits of Cross-ValidationStratified k-Fold Cross-Validation and Other StrategiesGrid SearchSimple Grid SearchThe Danger of Overfitting the Parameters and the Validation SetGrid Search with Cross-ValidationEvaluation Metrics and ScoringKeep the End Goal in MindMetrics for Binary ClassificationMetrics for Multiclass ClassificationRegression MetricsUsing Evaluation Metrics in Model SelectionSummary and Outlook6. Algorithm Chains and PipelinesParameter Selection with PreprocessingBuilding PipelinesUsing Pipelines in Grid SearchesThe General Pipeline InterfaceConvenient Pipeline Creation with make_pipelineAccessing Step AttributesAccessing Attributes in a Grid-Searched PipelineGrid-Searching Preprocessing Steps and Model ParametersGrid-Searching Which Model To UseSummary and Outlook7. Working with Text DataTypes of Data Represented as StringsExample Application: Sentiment Analysis of Movie ReviewsRepresenting Text Data as a Bag of WordsApplying Bag-of-Words to a Toy DatasetBag-of-Words for Movie ReviewsStopwordsRescaling the Data with tf-idfInvestigating Model CoefficientsBag-of-Words with More Than One Word (n-Grams)Advanced Tokenization, Stemming, and LemmatizationTopic Modeling and Document ClusteringLatent Dirichlet AllocationSummary and Outlook8. Wrapping UpApproaching a Machine Learning ProblemHumans in the LoopFrom Prototype to ProductionTesting Production SystemsBuilding Your Own EstimatorWhere to Go from HereTheoryOther Machine Learning Frameworks and PackagesRanking, Recommender Systems, and Other Kinds of LearningProbabilistic Modeling, Inference, and Probabilistic ProgrammingNeural NetworksScaling to Larger DatasetsHoning Your SkillsConclusionIndex
【深度学习】