Andy Eschbacher

Presentations

Talks, demos, and workshops given as a data scientist and educator

Me on the rim of Waimea Canyon

Talks

A sample of talks given at conferences, annual meetings, and meetups. Materials and recordings are given when available.

Upcoming

ODSC East - Boston, MA, USA 02 May 2019
Maps & Machine Learning: Best Practices for Great Cartography in Data Science
Making good maps in data science is hard. Maps are another form of data visualization with its own set of design principles to bring out the story in the data. Data scientists need for maps varies, but a very common use case is to quickly generate a data visualization iteratively to better understand model outputs, spatial variations in the data, and more.
In this mini-tutorial, I will cover best practices for data scientists needing maps -- ideas pulled from cartography and honed over the years working in the intersection of open source GIS and data science. Using open source tools, I will show how to make patterns more apparent, how to build interactive maps so that it is easier to explore data on larger and smaller scales, and list some cartographic tips and tricks. The goal is that data viz with maps will be made easier so that you can think about your data science problems instead of worrying about visualizing data on a map.

Past

Data Council - San Francisco, CA, USA 17 Apr 2019
Adding Spatial Structure and Data to Machine Learning Models (slide deck)
Spatial data science uses many of the same techniques and algorithm as traditional data science, but the spatial component can add a large amount of additional information by combining with other sources at the same location (e.g., census, geolocated tweets), using realtime routing services, or using the spatial structure of the distribution of the data.
In this talk, I will highlight work we have done with constraint-based clustering, probabilistic principal component analysis, and vanilla random forest that take special advantage of the spatial part of the data. I will further show how we build models from a variety of sources, including mobility data, points of interest, and accurate routing data.
For example, using the Python package CVXOPT, we solved a linear optimization problem that optimally distributes an asset from a source to a drain according to the road network and constraints that the drains cannot be over capacity, occasionally have fixed assignments, and all the asset has to be moved.
CARTO webinar - Internet 10 Jan 2019
Best Practices for Spatial Data Science using CARTO and Python (slide deck)
Learn how you can combine CARTO and Python for spatial data science from the comfort of your own Jupyter notebook. In this technical webinar, Andy Eschbacher (Senior Data Scientist at CARTO) and Joe Pringle (VP - North America at CARTO) will show how to apply CARTOframes and CARTO's Python SDK to build powerful end-to-end spatial data science workflows.
ODSC East - Boston, Mass, USA 01 May 2018
Spatial Data Science Analyses
Spatial data science uses many of the same techniques and algorithm as traditional data science, but the spatial component can add a large amount of additional information by combining with other sources at the same location (e.g., census, geolocated tweets), using realtime routing services, or using the spatial structure of the distribution of the data.

In this talk, I will highlight work we have done in linear optimization, genetic algorithms, and constraint-based clustering that take specital advantage of the spatial part of the data. For example, using the Python package CVXOPT, we solved a linear optimization problem that optimally distributes an asset from a source to a drain according to the road network and constraints that the drains cannot be over capacity, occasionally have fixed assignments, and all the asset has to be moved.
CARTO Locations - Madrid, Spain 20 Apr 2018
Blackbelt in spatial analytics
CARTO Locations - Madrid, Spain 19 Apr 2018
When Data Science Met LI: A Love Story
The birth of cartoframes as a project on the research team at CARTO and the need for better tools that fit in the data scientists' workflows. This talk will go beyond the origins of cartoframes and introduce the newest features and where were are going inf the future with this package.
ODSC Europe - London, England 14 Oct 2017
Open Spatial Data Science (slide deck)
The spatial part of a dataset gives more than just a lat/long. It allows you to thread a needle through any other spatial dataset that exists at that location: census, GPS tracks, data from a municipality's open data portal, and so much more. The spatial part is a key to a multidimensional world.

Augmenting your spatial data is only one piece, though. Once you know a location, you can use the measurements at the locations around you by appealing to Tobler's First Law of geography: "Everything is related to everything else, but near things are more related than distance things." Using this, statistics of geography allow you to find spatial correlations (Moran's I), calculate spatial regression (geographically weighted regression), and uncover spatial outliers (Getis-Ord's G*). At CARTO we're building these powerful techniques into an API (https://github.com/CartoDB/crankshaft) where data scientists can extract more value from his/her spatial data. Combined with the data augmentation process that we call the Data Observatory (https://carto.com/data-observatory), data scientists are freer and more enabled to explore their data in the context of the world.
Nacis Annual Meeting - Montréal, Canada 11 Oct 2017
Programmatic cartography in cartoframes (slide deck)
Discussion of the cartographic techniques built into cartoframes to help data scientist create better cartographic outputs based on the type of data they are visualizing
FOSS4G - Boston, MA 18 Aug 2017
Update on CARTO's Spatial Analytics Extension crankshaft (slide deck)
Discussion of the current state of crankshaft, CARTO's spatial analytic extension to PostgreSQL
Nacis Annual Meeting - Colorado Springs, CO 20 Oct 2016
Teaching New Cartography, with Rich Donohue (slide deck)
Hacks/Hackers NYC - New York City, NY 06 Sep 2016
Demo CARTO's Data Observatory that I helped build (slide deck)
NY Open Statistical Programming Meetup - New York, NY 11 Jul 2016
How the L Train Closure Affected Open Source Development of CartoDB (slide deck)
Discussing the work my team did on the Looking at the L project.
NYC Salon - New York, New York 10 Jul 2016
The Canarsie Closure and Carto: How a news leak altered the course of Carto's Development (slide deck)
Discussion about my team's work on the L Train closure. Talk started at 8th Ave L train and we looked at damage to the tunnel along the way. Once in Bushwick, I presented more in depth about our work and participants designed their own mitigation plan for what will happen for the 18 months the closure will be in effect.
PGConf US - Brooklyn, NY 18 Apr 2016
Taking PostgreSQL and Analytics to the Next Level with Python (slide deck)
Work using PL/Python to leverage the Python ecosystem with analysis that uses data stored in PostgreSQL
Texas GIS Forum - Austin, Texas 22 Oct 2015
Teaching in an Open Source World (slide deck)
Presentation of my philosophy and work on education in the classroom as a physics educator and then at a startup that uses web mapping technologies
FOSS4G - Seoul, South Korea 17 Sep 2015
Temporal Maps leading to new views in Spatial Analysis (slide deck)
Hacks/Hackers NYC - New York City, NY 11 Aug 2015
Building a map thinking machine: Demo One-click Mapping algorithm that I wrote (slide deck)
Bar Camp Philly - Philadelphia, PA 08 May 2015
Maps for Telling News Stories (slide deck)
I share some of my favorite recent maps used for contextualizing the news.
FOSS4G US - San Francisco, CA 10 Mar 2015
Everyone's a Geographer (slide deck)
Discussing my work developing open source geospatial curriculum for CARTO's Map Academy

Workshops and Presentations

I have given more workshops than I can remember, so what is below is a small sample of the ones that live online and are diverse in content.

Overview of CARTOframes 20 Apr 2018
Event listing
Venue CARTO Locations, Madrid, Spain
Skills Python, cartoframes, data science
Spatial Data Science 07 Feb 2017
Writeup, Event listing
Training an upcoming cohort of Data Scientists in spatial analysis and tools that can be integrated into the existing ecosystem of data science tools
Venue NYC Data Science Academy, New York, NY
Skills Python, cartoframes, data analysis, SQL, GIS
Data Science with Maps 27 Jan 2017
Writeup, Event listing
This workshop laid some of the groundwork for my furture work creating cartoframes.
Venue CartoCamp NYC, Brooklyn, NY
Skills Python, cartography, cartoframes, SQL
Explore Time and Space with SQL and PostGIS 16 Dec 2016
Writeup, Event listing
A workshop exploring PostgreSQL's date and PostGIS's spatial power to explore New York City through Taxi Data. We monitor taxi traffic to and from buildings and are able to identify, for example, which airport terminals and the times people boarded a taxi.
Venue CartoCamp NYC, Brooklyn, NY
Skills SQL, data analysis, data visualization, PostGIS
Visualizing Spatial Data on the Web 08 Dec 2016
Writeup, Event listing
Venue TechChange: Technology for Data Visualization, Online
Skills cartography, data visualization
Intermediate SQL Workshop 12 Jan 2016
Writeup, Event listing
Venue CartoCamp NYC, Brooklyn, NY
Skills SQL, data
Election Mapping Bonanza 10 Nov 2015
Writeup, Event listing
Venue CartoCamp NYC, Brooklyn, NY
Skills cartography, SQL

CartoCamp

I run a meetup called CartoCamp that teaches skills tangential to web mapping and spatial data science.