Data Analytics: A Small Data Approach is suitable for an introductory data analytics course to help students understand some main statistical learning models. It has many small datasets to guide students to work out pencil solutions of the models and then compare with results obtained from established R packages. Also, as data science practice is a process that should be told as a story, in this book there are many course materials about exploratory data analysis, residual analysis, and flowcharts to develop and validate models and data pipelines.
The main models covered in this book include linear regression, logistic regression, tree models and random forests, ensemble learning, sparse learning, principal component analysis, kernel methods including the support vector machine and kernel regression, and deep learning. Each chapter introduces two or three techniques. For each technique, the book highlights the intuition and rationale first, then shows how mathematics is used to articulate the intuition and formulate the learning problem. R is used to implement the techniques on both simulated and real-world dataset. Python code is also available at the book’s website: http://dataanalyticsbook.info.
Table of Contents
Who will benefit from this book
Overview of a Data Analytics Pipeline
Topics in a Nutshell
Regression & tree models
Logistic regression & ranking
Logistic Regression Model
A Ranking Problem by Pairwise Comparison
Statistical Process Control using Decision Tree
Bootstrap & random forests
How Bootstrap Works
5. LEARNING (I)
Cross validation & OOB
Out-of-bag error in Random Forest
Residuals & heterogeneity
Diagnosis in Regression
Diagnosis in Random Forests
7. LEARNING (II)
SVM & ensemble Learning
Support Vector Machine
LASSO & PCA
Principal Component Analysis
Experience & experimental
Kernel Regression Model
Conditional Variance Regression Model
Architecture & pipeline
APPENDIX: A BRIEF REVIEW OF BACKGROUND KNOWLEDGE
The normal distribution
Shuai Huang is an associate professor at the department of industrial & systems engineering at the university of Washington. He conducts interdisciplinary research in machine learning, data analytics, and applied operations research with applications on healthcare, manufacturing, and transportation areas.
Houtao Deng is a data science researcher and practitioner. He developed several new decision tree methods such as inTrees. He has built data-driven products for forecasting, scheduling, pricing, recommendation, fraud detection, and image recognition.