I try to read a few math books every year, but the more I learn, the more overlap I see in their content. In 2024 I decided to organize with a group of friends and read papers instead.

We read 17 papers together, and I skimmed many more to decide on which 17 to read in the group. Below is the list of papers, which might be useful if you want to learn some new data science related conceps, or if you want you organize a reading club of your own.

Lessons learned

Here’s what I learned:

• Most papers are not ideal for learning; the authors are not great writers and they have limited space to convey their ideas.
• Most practitioners should read books rather than papers to gain general knowledge.
• Finding papers that are on an appropriate technical level and of interest to everyone in a group is difficult.

Before reading papers, it might be useful to read some guides on reading papers:

• How to Read a Paper (2007)
• How to Read a Technical Paper (2009)
• A guide to reading scientific papers  (2016)
• How to read Machine Learning and Deep Learning Research papers (2021)

Here is the list of papers that we read. Honorable mentions are papers that I was interested in, but ended up not choosing for the group. Papers that I for some reason enjoyed more than others get a star (⭐).

Data Science

We chose to start with two papers on the culture of data science and practical tips for data science projects.

• 50 Years of Data Science. David Donoho (2017)
• ⭐ Good enough practices in scientific computing. Wilson et al. (2017)

Honorable mentions

• A few useful things to know about machine learning. Pedro Domingos (2012)
• ⭐ Statistical Modeling: The Two Cultures. Leo Breiman (2001)

Recommendation engines

We looked into recommendation engines: the Netflix price, matrix factorization and extensions to neural networks.

• Matrix Factorization Techniques for Recommender Systems. Koren et al. (2009)
• ⭐ Factorization Machines. Steffen Rendle (2010)
• Neural Collaborative Filtering. He et al. (2017) [arXiv]

Honorable mentions

• Collaborative Filtering for Implicit Feedback Datasets. Hu et al. (2008)
• Neural Factorization Machines for Sparse Predictive Analytics. He et al. (2017)

Continuous optimization

These articles cover practical methods for solving optimization problems.

• An overview of gradient descent optimization algorithms. Sebastian Ruder (2016)
• A comparison of numerical optimizers for logistic regression. Thomas P. Minka (2003)
• An Interior-Point Method for Large-Scale l1-Regularized Least Squares. Kim et al. (2007)

Honorable mentions

• Enhancing Sparsity by Reweighted L1 Minimization. Candes et al. (2007)

Operations research

In the age of data science, operations research is underrated.

• UPS Optimizes Delivery Routes. Holland et al. (2017)
• Increasing the Responsiveness of Firefighter Services …. van den Berg et al. (2017)
• Data analysis and optimization for (citi)bike sharing. O’Mahony et al. (2015)

Honorable mentions

• Defending Critical Infrastructure. Brown et al. (2006)

Information and compression

Information theory has a rich history and is the foundation of many modern techniques.

• A Mathematical Theory of Communication. Claude E. Shannon (1948)
• Arithmetic coding for data compression. Witten et al. (1987)

Statistics

More statistics never hurts, and seeing how practitioners attack problems is always interesting.

• Should the Olympic sprint skaters run 500 meter twice?. Nils Lid Hjort (1994)
• ⭐ A Conceptual Introduction to Hamiltonian Monte Carlo. Michael Betancourt (2017)
• MCMC using Hamiltonian dynamics. Radford M. Neal (2012)
• ⭐ Hierarchical generalized additive models in ecology. Pedersen​ et al. (2019)

Honorable mentions

• Mindless statistics. Gerd Gigerenzer (2004)
• Prediction, Estimation, and Attribution. Bradley Efron (2020)
• Hyperparameters: Optimize, or Integrate Out? David J.C. MacKay (1996)