book

Data Science Projects with Python - Second Edition

by Stephen Klosterman

July 2021

Beginner to intermediate

432 pages

9h 16m

English

Packt Publishing

Read now

Unlock full access

second edition
About the BookAbout the AuthorObjectivesAudienceApproachAbout the ChaptersHardware RequirementsSoftware RequirementsInstallation and SetupCode BundleAnaconda and Setting up Your EnvironmentConventionsCode PresentationGet in TouchPlease Leave a Review
IntroductionPython and the Anaconda Package Management SystemIndexing and the Slice OperatorExercise 1.01: Examining Anaconda and Getting Familiar with PythonDifferent Types of Data Science ProblemsLoading the Case Study Data with Jupyter and pandasExercise 1.02: Loading the Case Study Data in a Jupyter NotebookGetting Familiar with Data and Performing Data CleaningThe Business ProblemData Exploration Steps Exercise 1.03: Verifying Basic Data IntegrityBoolean MasksExercise 1.04: Continuing Verification of Data IntegrityExercise 1.05: Exploring and Cleaning the Data Data Quality Assurance and ExplorationExercise 1.06: Exploring the Credit Limit and Demographic FeaturesDeep Dive: Categorical FeaturesExercise 1.07: Implementing OHE for a Categorical FeatureExploring the Financial History Features in the DatasetActivity 1.01: Exploring the Remaining Financial Features in the DatasetSummary
IntroductionExploring the Response Variable and Concluding the Initial ExplorationIntroduction to Scikit-LearnGenerating Synthetic DataData for Linear RegressionExercise 2.01: Linear Regression in Scikit-LearnModel Performance Metrics for Binary ClassificationSplitting the Data: Training and Test SetsClassification AccuracyTrue Positive Rate, False Positive Rate, and Confusion MatrixExercise 2.02: Calculating the True and False Positive and Negative Rates and Confusion Matrix in PythonDiscovering Predicted Probabilities: How Does Logistic Regression Make Predictions?Exercise 2.03: Obtaining Predicted Probabilities from a Trained Logistic Regression ModelThe Receiver Operating Characteristic (ROC) CurvePrecisionActivity 2.01: Performing Logistic Regression with a New Feature and Creating a Precision-Recall CurveSummary
IntroductionExamining the Relationships Between Features and the Response VariablePearson CorrelationMathematics of Linear CorrelationF-testExercise 3.01: F-test and Univariate Feature SelectionFiner Points of the F-test: Equivalence to the t-test for Two Classes and CautionsHypotheses and Next StepsExercise 3.02: Visualizing the Relationship Between the Features and Response VariableUnivariate Feature Selection: What it Does and Doesn't DoUnderstanding Logistic Regression and the Sigmoid Function Using Function Syntax in PythonExercise 3.03: Plotting the Sigmoid FunctionScope of FunctionsWhy Is Logistic Regression Considered a Linear Model?Exercise 3.04: Examining the Appropriateness of Features for Logistic RegressionFrom Logistic Regression Coefficients to Predictions Using SigmoidExercise 3.05: Linear Decision Boundary of Logistic RegressionActivity 3.01: Fitting a Logistic Regression Model and Directly Using the CoefficientsSummary
IntroductionEstimating the Coefficients and Intercepts of Logistic RegressionGradient Descent to Find Optimal Parameter ValuesExercise 4.01: Using Gradient Descent to Minimize a Cost FunctionAssumptions of Logistic RegressionThe Motivation for Regularization: The Bias-Variance Trade-OffExercise 4.02: Generating and Modeling Synthetic Classification Data Lasso (L1) and Ridge (L2) RegularizationCross-Validation: Choosing the Regularization Parameter Exercise 4.03: Reducing Overfitting on the Synthetic Data Classification ProblemOptions for Logistic Regression in Scikit-LearnScaling Data, Pipelines, and Interaction Features in Scikit-LearnActivity 4.01: Cross-Validation and Feature Engineering with the Case Study DataSummary
IntroductionDecision TreesThe Terminology of Decision Trees and Connections to Machine LearningExercise 5.01: A Decision Tree in Scikit-LearnTraining Decision Trees: Node ImpurityFeatures Used for the First Splits: Connections to Univariate Feature Selection and InteractionsTraining Decision Trees: A Greedy AlgorithmTraining Decision Trees: Different Stopping Criteria and Other OptionsUsing Decision Trees: Advantages and Predicted ProbabilitiesA More Convenient Approach to Cross-ValidationExercise 5.02: Finding Optimal Hyperparameters for a Decision TreeRandom Forests: Ensembles of Decision TreesRandom Forest: Predictions and InterpretabilityExercise 5.03: Fitting a Random ForestCheckerboard GraphActivity 5.01: Cross-Validation Grid Search with Random ForestSummary
IntroductionGradient Boosting and XGBoostWhat Is Boosting?Gradient Boosting and XGBoostXGBoost HyperparametersEarly StoppingTuning the Learning RateOther Important Hyperparameters in XGBoostExercise 6.01: Randomized Grid Search for Tuning XGBoost HyperparametersAnother Way of Growing Trees: XGBoost's grow_policyExplaining Model Predictions with SHAP ValuesExercise 6.02: Plotting SHAP Interactions, Feature Importance, and Reconstructing Predicted Probabilities from SHAP ValuesMissing DataSaving Python Variables to a FileActivity 6.01: Modeling the Case Study Data with XGBoost and Explaining the Model with SHAPSummary
IntroductionReview of Modeling ResultsFeature EngineeringEnsembling Multiple ModelsDifferent Modeling TechniquesBalancing ClassesModel Performance on the Test SetDistribution of Predicted Probability and Decile ChartExercise 7.01: Equal-Interval ChartCalibration of Predicted ProbabilitiesFinancial AnalysisFinancial Conversation with the ClientExercise 7.02: Characterizing Costs and SavingsActivity 7.01: Deriving Financial InsightsFinal Thoughts on Delivering a Predictive Model to the ClientModel MonitoringEthics in Predictive ModelingSummary
1. Data Exploration and CleaningActivity 1.01: Exploring the Remaining Financial Features in the Dataset2. Introduction to Scikit-Learn and Model EvaluationActivity 2.01: Performing Logistic Regression with a New Feature and Creating a Precision-Recall Curve3. Details of Logistic Regression and Feature ExplorationActivity 3.01: Fitting a Logistic Regression Model and Directly Using the Coefficients4. The Bias-Variance Trade-OffActivity 4.01: Cross-Validation and Feature Engineering with the Case Study Data5. Decision Trees and Random ForestsActivity 5.01: Cross-Validation Grid Search with Random Forest6. Gradient Boosting, XGBoost, and SHAP ValuesActivity 6.01: Modeling the Case Study Data with XGBoost and Explaining the Model with SHAP 7. Test Set Analysis, Financial Insights, and Delivery to the ClientActivity 7.01: Deriving Financial Insights

Overview

Data Science Projects with Python offers a hands-on, project-based approach to learning data science using real-world data sets and tools. You will explore data using Python libraries like pandas and Matplotlib, build machine learning models with scikit-learn, and apply advanced techniques like XGBoost and SHAP values. This book equips you to confidently extract insights, evaluate models, and deliver results with clarity.

What this Book will help me do

Learn to load, clean, and preprocess data using Python and pandas.
Build and evaluate predictive models, including logistic regression and random forests.
Visualize data effectively using Python libraries like Matplotlib.
Master advanced techniques like XGBoost and algorithmic fairness.
Communicate data-driven insights to aid decision making in practical scenarios.

Author(s)

Stephen Klosterman is an experienced data scientist with a strong focus on practical applications of machine learning in business. Combining a rich academic background with hands-on industry experience, he excels at explaining complex concepts in an approachable way. As the author of 'Data Science Projects with Python,' his goal is to provide learners with the skills needed for real-world data science challenges.

Who is it for?

This book is ideal for beginners in data science and machine learning who have some basic programming knowledge in Python. Aspiring data scientists will benefit from its practical, end-to-end examples. Professionals seeking to expand their skillset in predictive modeling and delivering business insights will find this book invaluable. Some foundation in statistics and programming is recommended.