PyData Amsterdam 2018
2018-05-25..2018-05-27
Amsterdam, Netherlands
[PyData Amsterdam 2016, PyData Amsterdam 2017, PyData Amsterdam 2018, PyData Amsterdam 2019, PyData Amsterdam 2024, PyData Amsterdam 2025]
| title | author | length |
|---|---|---|
| PyData Amsterdam 2018 | 03m | |
Conference highlights include a deployed air pollution web and mobile app that publishes Macedonian sensor readings and helps trigger a 3,000-strong protest… | ||
| Keynote: Are You Ready for GDPR? | Michele Appello | 47m |
A practical checklist: a published GDPR compliance primer plus seven concrete steps — audit every data flow in a spreadsheet, color‑code risk, run data… | ||
| Keynote: Pentesting ChatOps | Melanie Rieback | 49m |
Claims that ChatOps makes pen testing repeatable and transparent and shows the tooling and results: self‑hosted Rocket.Chat plus a Hubot‑style bot called… | ||
| Hands-on introduction to Deep Learning with Keras and Tensorflow | Rodrigo Agundez | 01h 29m |
Hands-on notebooks show building models with Keras with Tesorflow backend, training dense and convolutional nets, and using batch normalization and dropout… | ||
| Structured Streaming with PySpark and Azure Databricks | Andrei Varanovich | 01h 21m |
Structured streaming claims to make real‑time “continuous applications” practical by treating streams as tables and keeping computations consistent; the demo… | ||
| Conversational AI with Rasa Core & NLU | Tom Bocklisch | 01h 38m |
Live-coding builds a working ML chatbot with Rasa NLU & Rasa Core, using spaCy medium vectors for intent features, CRF for entities and an LSTM policy for… | ||
| Tricks, tips and topics in Text Analysis | Bhargav Srinivasa Desikan | 01h 29m |
Shows an actual pipeline: use the spaCy pipeline to tokenize, POS‑tag, lemmatize and mark stopwords, build bigrams with gensim, make a gensim dictionary/corpus… | ||
| Creating correct and capable classifiers | Ian Ozsvald | 37m |
A practical workflow that starts with a concrete baseline and ends with explainable models the talk claims reduces guesswork and reveals real bugs: use… | ||
| Forecasting Airline Passengers Using Designer Machine Learning | Alexander Backus, Jan van der Vegt | 34m |
Forecasts passenger counts for individual flights using a progression of models from a simple linear regression to feed-forward and recurrent deep neural… | ||
| Detecting Pikachu on Android using Tensorflow Object Detection | Juan De Dios Sanos Rivera | 34m |
Builds a Pikachu detector with TensorFlow Object Detection, converts labeled XML to TFRecord and trains an SSD + MobileNet pipeline on ~230 images with an… | ||
| SAMENVATTR, or, How to Make a Text Summarization Tool for your Language of Choice | Robert Rodger | 36m |
Extractive summarization uses text rank by building a sentence graph, weighting edges with BM25/TF‑IDF after stop‑word removal and stemming, and solving a… | ||
| The Data I Breathe | Gorjan Jovanovski | 26m |
Open datasets become a live civic tool: raw government air readings + a web and mobile app turn garbled JSON into clear visuals, reach 1,000,000 users within… | ||
| Super Donkey Cars | Niels Zeilemaker | 33m |
An open source Donkey car is rebuilt to train on an AWS GPU instance and then run real-time inference on a Raspberry Pi; hardware costs ~€220, data collection… | ||
| Selection bias: The elephant in the room | Lucas Javier Bernardi | 37m |
Problem: training labels come only from later funnel stages so measured gains don’t reflect true production performance; experiments with a family‑trip… | ||
| Learning credit networks with applications in financial industry | Sumit Sourabh | 38m |
A concrete method learns a Bayesian credit network from CDS spreads by detecting epsilon drops (thresholded jump points) and fitting structure and parameters… | ||
| Thompson Sampling for Machine Learning | Ruben Mak | 35m |
Overview of applying Thompson Sampling to machine learning problems, including technical examples like using Bayesian Neural Networks implemented with Edward… | ||
| Winning (Some) Churners Back | Ferran Elias | 30m |
Booking.com builds a churn-score pipeline and an experimentation framework that targets incentives to users by score rather than blasting everyone; early model… | ||
| Why AI in the Medical Field Both Sucks and Rocks... | Kicky van Leeuwen | 34m |
Case study on segmentation of 4D heart MRI uses a U-net style model trained in tensor flow Cara's Google cloud platform on Philips scans where one patient… | ||
| Improve classification accuracy for customer contact transcriptions | Maria Vechtomova | 31m |
KPN processes ~7 million calls and ~700k chats yearly and starts from 180k labeled calls and 40k chats to compare Naive Bayes, CNN with word embeddings (and… | ||
| Monitoring models in production | Jannes Klaas | 26m |
Monitors Keras neural nets in production, shows a concrete pipeline: serve predictions, send inputs and outputs to a separate watchman instance that stores… | ||
| Deploying JupyterHub with Kubernetes | Camilla Montonen | 27m |
Shows how a production JupyterHub is built on Kubernetes with concrete choices: GKE + Helm chart for quick deploy, alternatives on Azure and DIY on AWS, and a… | ||
| Unit Testing Data with Marbles | Jane Stewart Adams | 40m |
Argues that unit testing is a practical framework for validating data assumptions and shows a working Python extension that does it: Two Sigma's Marbles… | ||
| Computer-aided design of advertisement product compositions | Kay Hoogland | 25m |
Claims semi-automated creation of product compositions by letting designers steer an interactive genetic algorithm that generates layouts and applies visual… | ||
| Elegant data pipelining with Apache Airflow | Bolke de Bruin | 36m |
Claims that pipelines must be idempotent and deterministic and use immutable partitions guide the concrete examples: Airflow templates execution_date, defines… | ||
| Case Study in Travel Business - Time Series Analysis with Seasonal Data | Cheuk Ting Ho | 25m |
Case study on B2B worldwide travel bookings shows a concrete pipeline: decompose series into trend seasonality and noise using the open source Seasonal library… | ||
| Half Hour of Flower: Planting the seeds for great Image Quality | Dirk Guijt | 34m |
Royal FloraHolland builds a fully automated self‑improving image‑quality pipeline that uses VGG16 transfer learning for per‑property classifiers, wraps models… | ||
| (Re)training word embeddings for a specific domain | Jetze Schuurmans | 30m |
Claims that retraining GloVe / word2vec / fastText on a domain corpus improves domain terms and senses, then shows concrete steps: build a Wikipedia domain… | ||
| Scaling Machine Learning jobs with Kubernetes | Tarek Mehrez, Carsten Lygteskov Hansen | 26m |
Textkernel runs TensorFlow training on a fixed on‑prem Kubernetes cluster with Docker images built by Jenkins from GitLab, using pod anti‑affinity so each… | ||
| Python and the Holy Grail of Causal Inference | Dennis Ramondt, Huib Keemink | 31m |
Claims that ruined randomized tests can be fixed with observed covariates, demonstrated by a heat‑pump deployment where propensity score matching corrects… | ||
| DataFrames: scaling up and out | Ondrej Kokes | 32m |
Pandas works in memory so the checklist is: avoid per-row Python loops and use vectorized ops to get 100x speedups; load only needed columns, use chunking or… | ||
| Building customer-visible data science dashboards with Altair / Vega / Vue | Uwe L. Korn | 29m |
Argues that shipping customer-facing dashboards requires a compromise: Python generates Vega/Vega‑Lite specs (Altair converts pandas data frames to JSON) while… | ||
| Bayesian Deep learning with 10% of the weights | Rob Romijnders | 36m |
Claims a variational Bayesian net that learns Gaussian mu and sigma per weight and samples weights at inference to get a posterior predictive, implemented in… | ||
| “Why Should I Trust You?” - Debugging black-box text classifiers | Tobias Sterbak | 23m |
Claims that eli5 and LIME produce faithful local explanations of black‑box text classifiers by training a simple surrogate (CountVectorizer + logistic… | ||
| From cells to drug responses - machine learning in cancer research | Julian de Ruiter, Nanne Aben | 35m |
Predicts drug response from tumor sequencing using Snakemake/Bioconda to build reproducible pipelines, fits simple linear models with elastic net and tandem… | ||
| Dynamic Code Instrumentation with Hacked Interpreters | James Powell | 38m |
Shows a working patched CPython that instruments every bytecode push by overriding the push macro, records when specific objects are touched, and emits JSON… | ||
| Challenges in building Machine Learning models in production | Elena Sokolova | 24m |
Practical account of deploying personalization on booking.com: concrete failure modes, fixes and measured effects. Covers continuous cold start and using… | ||
| Asyncio, websockets, and BTC sitting in a tree | Giovanni Lanzani | 37m |
A side project implements a real arbitrage detector that consumes exchange websocket feeds with asyncio + websockets, inserts ticks into Postgres via asyncpg +… | ||
| Lightning Talks | 24m | |
A set of short demos shows concrete tools and results: a notebook diff that strips metadata and renders succinct comparisons so a 200‑cell file nbdime… | ||
| nothing matches | ||