Post-Shrinkage Strategies in Statistical and Machine Learning for High Dimensional Data

This book presents post-estimation and predictions strategies for statistical models with applications in data science, combining statistical learning and machine learning techniques. It suggests shrinkage strategies to control bias and can be implemented for high-dimensional data analysis. It is an excellent textbook for advanced undergraduate and graduate courses and will help researchers and practitioners develop improved estimation strategies.

Format: Hardback
Length: 378 pages
Publication date: 25 May 2023
Publisher: Taylor & Francis Ltd

This comprehensive book delves into advanced post-estimation and prediction strategies for a wide range of valuable statistical models with applications in data science. By seamlessly integrating statistical learning and machine learning techniques, it offers a unique and optimal approach. It is widely acknowledged that machine learning methods face numerous challenges related to bias, leading to potential issues such as exploding mean squared error and prediction error. To address this, the book proposes shrinkage strategies to mitigate bias by combining a submodel selected through a penalized method with a model featuring numerous features. Furthermore, the proposed shrinkage methodology proves effective for high-dimensional data analysis. Researchers in statistics and medical sciences, who frequently work with large datasets, rely on statistical modeling to analyze these data. Accurate estimation of model parameters is crucial in data analysis. This book serves as a valuable resource for developing improved estimation strategies for statisticians. It will be of immense benefit to researchers and practitioners for their teaching and advanced research, and serves as an excellent textbook for advanced undergraduate and graduate courses focusing on shrinkage, statistical, and machine learning.

The book adeptly highlights the bias inherent in machine learning methods and presents practical tools, techniques, and tips to address this issue. It sheds light on the fundamental reasoning behind model selection and post-estimation using shrinkage and related strategies, making it a foundational presentation in this rapidly evolving field. Shrinkage and other appropriate methods for model selection and estimation problems are gaining significant interest as researchers aim to bridge the gap between competitive strategies. By applying these strategies to real-life data, practitioners can gain valuable insights and make informed decisions based on empirical evidence.

Weight: 884g
Dimension: 184 x 261 x 28 (mm)
ISBN-13: 9780367763442