Building a Recommendation System with R

Written by  Suresh K. Gorakala and Michele Usuelli, published by Packt Press 2015

This is whole book on a topic that is often only a single chapter in a book. It is a book for people who already know R and machine learning .

The book uses Math equations not just code for teaching the concepts.

Covers confusion matrix for classification. Along with sensitivity and specification.  Lots of details about type one and type two errors. This  clearly written section will help you understand why you don’t want either type of error and what they are.

Classification similarity measures include Euclidean Distance, Cosine Distance and Pearson Correlation.

Dimensionality  reduction techniques include Principle Component Analysis.

Data Mining techniques include K-means clustering and Support Vector Machine.

Recommender System includes collaborative filtering and content based filtering.

 

R package for the book is recommenderlab.

recommenderlab: Lab for Developing and Testing Recommender Algorithms by Michael Hahsler at http://CRAN.R-project.org/package=recommenderlab

Other packages used are lsa, e1071, cluster.

 

On Meaningful Scientific Laws

On Meaningful Scientific Laws book cover
On Meaningful Scientific Laws

Written by Jean-Claude Falmange and Christopher Doble published by Springer 2015

This is a concise book full of proofs. They define Scientific Law in the Preface. An equation in which the variables represent quantities that are physical or geometrical. Meaningful is described in Chapter 5, Defining meaningfulness.

I read this book to take a break from the tangles data messes I am straightening out. It was a pleasant break.  This is a well written book with concise definitions. Well constructed proofs. I think a patient beginner could understand the book.

Useful ideas that would lead to not tangling data in the first place, like ratio scales in Chapter 2.

Chapter 9 Dimensional Invariance and Dimensional Analysis. You have to read thru the whole book because each section builds on previous to get to this valuable part. It is worth for the knowledge.

Now I am ready to get back to untangling data.