Over the years, I have used a number of books for Econometrics, statistics and machine learning either as a student, teaching assistant or lecturer. Even though I enjoy reading hard copies more, I am more and more inclined to have electronic copies when overall costs of carry are considered. Given that more and more electronic books can be freely accessible online, I feel that there is little incentive to have a hard copy if you do not already own one. Below is a list of openly accessible books that I find helpful especially for people interested in econometrics, statistics, and machine learning.
Econometrics
Statistics
Machine Learning
The Elements of Statistical Learning: Data Mining, Inference, and Prediction
Statistical Learning with Sparsity: The Lasso and Generalizations
An Introduction to Statistical Learning with Applications in R
The books that I still keep hard copies with me are Econometric Analysis by William H. Greene, Time Series Analysis by James D. Hamilton and Introductory Econometrics by Jeffrey M. Wooldridge. The first two are great reference books for econometric theory and the last one comes very handy when looking for small economic data sets.