Statistics with Rust: 50+ Statistical Techniques Put into Action

Statistics with Rust

50+ Statistical Techniques Put into Action

Keiko Nakamura

Copyright © 2023 by GitforGits.

All rights reserved. This book is protected under copyright laws and no part of it may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without the prior written permission of the publisher. Any unauthorized reproduction, distribution, or transmission of this work may result in civil and criminal penalties and will be dealt with in the respective jurisdiction at anywhere in India, in accordance with the applicable copyright laws.

Published by: GitforGits

Publisher: Sonal Dhandre

www.gitforgits.com

[email protected]

Printed in India

First Printing: April 2023

Cover Design by: Kitten Publishing

For permission to use material from this book, please contact GitforGits at [email protected].

Content

Preface

Chapter 1: Introduction to Rust for Statisticians

Why Rust for Data Analysis and Statistics?

Comparing Rust and Python for Statistics

Performance

Memory Safety and Resource Management

Concurrency

Interoperability

Ecosystem Growth and Future Prospects

Readability and Maintainability

Scalability

Cross-platform and Deployment

Learning Curve

Setting up Rust Environment

Download rustup-init

Run rustup-init

Configure PATH Environment Variable

Verify the Installation

Essential Rust Libraries for Statistics

ndarray

statrs

statis

plotly

Setting up Statistical Project

Create a New Rust Project

Add Library Dependencies

Build and Run the Project

Import the Libraries in Rust Code

Summary

Chapter 2: Data Handling and Preprocessing

Data Handling and Preprocessing

Process of Data Handling and Preprocessing

Exploring CSV crate

Dataset Loading with CSV crate

Parsing the Data

Data Structures in Rust

Arrays

Vectors

Tuples

Structs

HashMaps

Calculating Mean

Calculating Median

Common Data Cleaning and Preprocessing Techniques

Handling Missing Values

Data Type Conversion

Scaling/Normalizing Data

Encoding Categorical Variables

Feature Engineering

Performing Data Cleaning and Preprocessing

Summary

Chapter 3: Descriptive Statistics in Rust

Introduction to Descriptive Statistics

Measures of Central Tendency

Calculate Measures of Central Tendency

Measures of Dispersion

Calculate Measures of Dispersion

Exploratory Data Analysis (EDA)

Implementing EDA

Summary

Chapter 4: Probability Distributions and Random Variables

Discrete Probability Distribution

Uniform Distribution

Bernoulli Distribution

Binomial Distribution

Poisson Distribution

Geometric Distribution

Continuous Probability Distribution

Uniform Distribution

Normal (Gaussian) Distribution

Exponential Distribution

Beta Distribution

Gamma Distribution

Generating Random Variables

Sampling from Distributions

Sample Program for Sampling from Distributions

Estimating Distribution Parameters

Method of Moments (MoM)

Maximum Likelihood Estimation (MLE)

Bayesian Estimation

Least Squares

Summary

Chapter 5: Inferential Statistics

Fundamentals of Inferential Statistics

Hypothesis Testing

Confidence Intervals

Performing Hypothesis Testing

Two-sample T-test

Chi-square Test for Independence

Calculating Confidence Interval

For Mean

For the Proportion

Parametric Tests

Paired T-test

One-way ANOVA

Non-parametric Tests

Wilcoxon Rank-sum Test (Mann-Whitney U Test)

Implementing Wilcoxon Rank-sum Test

Kruskal-Wallis Test

Implementing Kruskal-Wallis Test

Summary

Chapter 6: Regression Analysis

Introduction to Regression Analysis

Overview

Applications of Regression Analysis

Types of Regression Analysis

Simple Linear Regression

Understanding Equation

Applying Simple Regression with Rust

Multiple Linear Regression

Understanding Equation

Applying Multiple Linear Regression

Polynomial Regression

Understanding Equation

Applying Polynomial Regression

Ridge and Lasso Regression

Understanding Equation

Applying Ridge and Lasso Regression

Logistic Regression

Understanding Equation

Applying Logistic Regression

Summary

Chapter 7: Bayesian Statistics

Introduction to Bayesian Statistics

Bayes Theorem

Advantages of Bayesian Statistics

Bayesian Inference

Putting Bayesian Inference into Action

Procedure to Perform Bayesian Inference

Practical Illustration of Bayesian Inference

Bayesian Model Comparison

Bayesian Hierarchical Modeling

Advanced Markov Chain Monte Carlo Method

Simple Implementation of HMC Method

Model Comparison and Selection

Model Comparison using DIC

Model Comparison using WAIC

Summary

Chapter 8: Multivariate Statistical Methods

Multivariate Statistical Methods

Introduction

Overview of Multivariate Techniques

Principal Component Analysis (PCA)

Procedure of PCA

Sample Program to Implement PCA

Canonical Correlation Analysis (CCA)

Procedure to Perform CCA

Sample Program to Implement CCA

Linear Discriminant Analysis (LDA)

Procedure to Perform LDA Algorithm

Sample Program to Implement LDA

Independent Component Analysis (ICA)

Overview of ICA Algorithm

Sample Program to Implement ICA

Multidimensional Scaling (MDS)

Types of Multidimensional Scaling

Sample Program to Implement Classical MDS

Summary

Chapter 9: Nonlinear Models and Machine Learning

Nonlinear Models

Decision Trees

Overview

Building Decision Tree

Support Vector Machines (SVM)

Overview

Building SVM Model

Neural Networks

Fundamentals of Neural Networks

Building Neural Network Model

Ensemble Methods

Overview

Building Bagging Ensemble of Decision Tree

Summary

Chapter 10: Model Evaluation and Validation

Model Evaluation and Validation

Introduction

Train-test Split Technique

Exploring Train-test Split

Implementing Train-test Split

Cross-validation Technique

Understanding Cross-validation

Implementing K-fold Cross-validation

Hyperparameter Tuning

Overview

Perform Hyperparameter Tuning using Grid Search

Model Selection Techniques: AIC and BIC

Akaike Information Criterion (AIC)

Bayesian Information Criterion (BIC)

Implement AIC and BIC

Resampling Methods

Bootstrapping

Permutation Tests

Perform Bootstrapping and Permutation Test

Implementing Bootstrapping

Implementing Permutation Test

Summary

Chapter 11: Text and Natural Language Processing

Overview of Natural Language Processing (NLP)

Key Processes of NLP

Text Preprocessing and Tokenization

Key Preprocessing Techniques

Common Tokenization Approaches

Implementing Text Preprocessing and Tokenization

Sample Program to Perform Preprocessing and Tokenization

Stopword Removal Process

Sample Program to Perform Stopword Removal

Stemming and Lemmatization

Perform Stemming

Information Retrieval with TF-IDF

TF-IDF Components

Implementation of TF-IDF

Word Embeddings and Word2Vec

Summary

Index

Epilogue

Preface

Are you an experienced statistician or data professional looking for a powerful, efficient, and versatile programming language to turbocharge your data analysis and machine learning projects? Look no further! Statistics with Rust is your comprehensive resource to unlock Rust's true potential in modern statistical methods.

This book is tailored specifically for statisticians and data professionals who are already familiar with the fundamentals of statistics and want to leverage the speed and reliability of Rust in their projects. Over 11 in-depth chapters, you will discover how Rust outperforms Python in various aspects of data analysis and machine learning and learn to implement popular statistical methods using Rust's unique features and libraries.

Statistics with Rust begins by introducing you to Rust's programming environment and essential libraries for data professionals. You'll then dive into data handling, preprocessing, and visualization techniques that form the backbone of any statistical analysis. As you progress through the book, you'll explore descriptive and inferential statistics, probability distributions, regression analysis, time series analysis, Bayesian statistics, multivariate statistical methods, and nonlinear models. Additionally, the book covers essential machine-learning techniques, model evaluation and validation, natural language processing, and advanced techniques in emerging topics.

In this book you will learn how to:

Discover Rust's unique advantages for statistical analysis and machine learning projects.

Learn to efficiently handle, preprocess, and visualize data using Rust libraries.

Implement descriptive and inferential statistics with Rust for powerful data insights.

Master probability distributions and random variables in Rust for robust simulations.

Perform advanced regression analysis with Rust's capabilities.

Explore Bayesian statistics and Markov Chain Monte Carlo methods in Rust.

Uncover multivariate techniques, including PCA and Factor Analysis, using Rust libraries.

Implement cutting-edge machine learning algorithms and model evaluation techniques in Rust.

Delve into text analysis, and natural language processing with Rust.

To ensure you get the most out of this book, each chapter includes hands-on examples and exercises to reinforce your understanding of the concepts presented. You'll also learn to optimize your Rust code and select the best tools and libraries for each task, maximizing your productivity and efficiency.

GitforGits

Prerequisites

Statistics with Rust is your indispensable guide to harnessing the power of Rust for modern statistical analysis and machine learning. Whether you are a seasoned data professional or a Rust enthusiast looking to expand your knowledge, this book provides the tools and insights to elevate your projects.

Codes Usage

Are you in need of some helpful code examples to assist you in your programming and documentation? Look no further! Our book offers a wealth of supplemental material, including code examples and exercises.

Not only is this book here to aid you in getting your job done, but you have our permission to use the example code in your programs and documentation. However, please note that if you are reproducing a significant portion of the code, we do require you to contact us for permission.

But don't worry, using several chunks of code from this book in your program or answering a question by citing our book and quoting example code does not require permission. But if you do choose to give credit, an attribution typically includes the title, author, publisher, and ISBN. For example, Statistics with Rust by Keiko Nakamura.

If you are unsure whether your intended use of the code examples falls under fair use or the permissions outlined above, please do not hesitate to reach out to us at [email protected]. 

We are happy to assist and clarify any concerns.

Acknowledgement

I owe a tremendous debt of gratitude to GitforGits, for their unflagging enthusiasm and wise counsel throughout the entire process of writing this book. Their knowledge and careful editing helped make sure the piece was useful for people of all reading levels and comprehension skills. In addition, I'd like to thank everyone involved in the publishing process for their efforts in making this book a reality. Their efforts, from copyediting to advertising, made the project what it is today.

Finally, I'd like to express my gratitude to everyone who has shown me unconditional love and encouragement throughout my life. Their support was crucial to the completion of this book. I appreciate your help with this endeavour and your continued interest in my career.

Chapter 1: Introduction to Rust for Statisticians

Why Rust for Data Analysis and Statistics?

In recent years, the Rust programming language has attracted considerable attention from developers for its safety, speed, and concurrency capabilities. Originating as a systems programming language, Rust has grown in popularity and has been adopted across various domains, including web development, embedded systems, and even data analysis. With its focus on performance and safety, Rust is a formidable choice for data analysis and statistical computing, providing unique advantages over traditional languages such as Python, R, and Julia.

This book aims to guide you through the world of statistics and data analysis using Rust, offering a comprehensive understanding of Rust's potential in these fields. By the end of this journey, you will be equipped with the knowledge and practical skills to leverage Rust's power for your data analysis projects.

Rust's high-performance capabilities are one of its most appealing features. As a compiled language, Rust offers performance that is on par with or even surpasses C and C++. This is particularly important for data analysis and statistics, where large datasets and complex computations are common. With Rust, you can execute data processing tasks and run algorithms with lower latency, enabling faster and more efficient analysis.

Memory safety is a critical aspect of any programming language, especially when dealing with large datasets or complex data structures. Rust's unique ownership system and strong type system ensure memory safety at compile time, eliminating common bugs such as data races, null pointer dereferences, and buffer overflows. This guarantees that your data analysis programs will be more robust and less prone to crashes, without the need for a garbage collector that might impact performance.

Modern hardware often features multiple cores or processors, and utilizing this parallelism is essential for high-performance computing. Rust's built-in concurrency support, based on its ownership and borrowing system, allows you to build concurrent and parallel programs with ease. By leveraging Rust's concurrency features, you can efficiently distribute data processing tasks across multiple cores or even multiple machines, significantly reducing the time required for complex calculations.

Rust's C-compatible FFI (Foreign Function Interface) enables seamless integration with existing C and C++ libraries. This means you can easily use existing high-performance libraries for data analysis, such as BLAS (Basic Linear Algebra Subprograms), LAPACK (Linear Algebra PACKage), or FFTW (Fastest Fourier Transform in the West), alongside Rust's native libraries. Moreover, Rust's WebAssembly support allows you to run your data analysis code on the web, opening up new possibilities for interactive data visualization and analysis tools.

Although Rust is a relatively young language, its ecosystem has grown rapidly, with an ever-increasing number of libraries and tools catering to data analysis and statistics. Libraries such as ndarray, statrs, and plotly offer robust support for data manipulation, statistical computation, and visualization. Additionally, the Rust community is highly active and committed to developing new libraries and improving existing ones, ensuring that the Rust ecosystem will continue to expand and evolve.

Rust's syntax is clear, concise, and expressive, making it easier for you to write and read your code. This improves the maintainability of your data analysis programs, allowing you to