Szilard Pafka

benchm-ml. A minimal benchmark for scalability, speed and accuracy of commonly used open source implementations (R packages, Python scikit-learn, H2O, xgboost, Spark MLlib etc.) of the top machine learning algorithms for binary classification (random forests, gradient boosted trees, deep neural networks etc.).

GBM-perf. Performance of various open source GBM implementations

teach-data-science-UCLA-master-appl-stats. Materials for STATS 418 - Tools in Data Science course taught in the Master of Applied Statistics at UCLA

benchm-databases. A minimal benchmark of various tools (statistical software, databases etc.) for working with tabular data of moderately large sizes (interactive data analysis).

benchm-dl. Playing with various deep learning tools and network architectures

ml-prod. Some thoughts on how to use machine learning in production

survey-ml-tools. Quick informal survey at the Los Angeles Machine learning meetup about tools used for machine learning.

teach-data-science-msc-analytics-ceu. Materials for a short introductory/intermediate Data Science course taught in the MSc in Business Analytics program at the Central European University

xgboost-adv-workshop-LA. Advanced workshop on XGBoost with Tianqi Chen in Santa Monica, June 2, 2016

teach-ML-CEU-master-bizanalytics. Machine Learning #1 and #2 courses at CEU Master of Science in Business Analytics

GBM-tune. Tuning GBMs (hyperparameter tuning) and impact on out-of-sample predictions

datascience-latency. Latency numbers every data scientist should know (aka the pyramid of analytical tasks) - the order of magnitude of computational time for the most common analytical tasks (SQL-like data munging, linear and non-linear supervised learning etc.) with the typically available tools on commodity hardware.

GBM-multicore. GBM multicore scaling: h2o, xgboost and lightgbm on multicore and multi-socket systems

ML-scoring. Compare the scoring speed of several open source machine learning libraries.

GBM-intro. GBM intro talk (with R and Python code)

dataset-sizes-kdnuggets. Size of datasets used for analytics based on 10 years of surveys by KDnuggets.

talks-main. Most recent/important talks given at conferences/meetups

GBM-adv-workshop-Bp19. Advanced GBM Workshop - Budapest, Nov 2019

kaggle-scripts-R-pydata. Kaggle scripts: R vs pydata + most popular R and Python packages for Machine Learning

benchm-dplyr-dt.

awesome-GBMs. A curated list of gradient boosting machines (GBM) resources

datascience-course-historical. Inspired by David Donoho's "50 Years of Data Science" (2015) paper, I'm releasing here a course proposal draft I wrote in 2009 for a possible course of "data science".

dscomp-winstab. Winner stability in data science competitions

ml-algos-perf. Performance of Machine Learning Algorithms - playground for experimentation in order to understand their performance characteristics as a function of the attributes of the datasets used for training

h2o-scoring--OLD. Various options for deploying h2o.ai models to production (scoring new data)

GBM-workshop. Code (and other materials) for an introductory talk/workshop on GBMs (developed originally for an R-Ladies Meetup)

DS_meetups. Contents from the Real Data Science USA (formerly LA Data Science) Meetup

datascience-1slide. Data Science in 1 Slide

aboutme. HTML

student-data-science-project-2. Sample student project for the Data Science course I was teaching at CEU's MSc in Business Analytics https://github.com/szilard/teach-data-science-msc-analytics-ceu

ml-x1. Machine learning tools on monster EC2 X1 instance (128 cores, 2 TB RAM)

benchm-ml-talks.

shinyvalidinp. R

GBM-meltdown. The Effect of the Linux Kernel Page-Table Isolation (KPTI) Patch (Meltdown Vulnerability) on GBMs

bio. Szilard Pafka's short bio (to go with conference talk abstracts)

benchm-R-mysql. R

student-data-science-project-1-kaggle. Sample student project for the Data Science course I was teaching at CEU's MSc in Business Analytics https://github.com/szilard/teach-data-science-msc-analytics-ceu

LA-data-meetups.

BigDataDayLA2015-DataScience. List of talks from the Data Science Track of Big Data Day LA 2015 (annual free conference)

2018-n9-bRew-and-View. Come join us for the first ever R-themed Brew and View! We are going to curate a few of our favorite talks into 20 minute clips, and then highlight key points and discuss afterward.

benchm-ml--data. Data for benchm-ml, gbm-perf etc. (samples from the airline dataset)

app-consumer-loan. HTML