PyData Eindhoven 2022
2022-12-01..2022-12-03
Eindhoven, Netherlands
[PyData Eindhoven 2019, PyData Eindhoven 2020, PyData Eindhoven 2021, PyData Eindhoven 2022, PyData Eindhoven 2023, PyData Eindhoven 2024, PyData Eindhoven 2025]
who it's for
data scientists, machine learning engineers, data engineers, python developers, mlops engineers, analytics engineers
topics
DuckDB zero-copy Pandas/Arrow; DataFrame frameworks comparison; reproducible ML with DVC; Devcontainer for Python projects; ml monitoring and observability; causal inference with Temporal Fusion Transformer; position-bias corrected grocery ranking; Thompson sampling for ad personalization; fuzzy topic modelling in Python; bulk labelling techniques; notebook usability improvements and fixes; low-code deployment with Taipy
| title | author | length |
|---|---|---|
| Aftermovie | Gareth Thomas | 02m |
A fast cut montage of conference highlights shows people, booths and short demos rather than a technical lecture: quick clips of microPython boards running on… | ||
| Opening Notes / AI Ethics in the Wild - Welcome to the Jungle | Fabian vd Berg, Marc van Meel | 47m |
An on‑device deep‑learning image recognizer for a zoo badge app collects user photos in an automated feedback loop and unintentionally accepts souvenir and… | ||
| DuckDB: Bringing analytical SQL directly to your Python shell | Pedro Holanda | 30m |
DuckDB is an embedded analytical database (pre-release v0.6) that executes SQL without a server and offers zero-copy integration with Pandas and Arrow, a… | ||
| Causal inference and scenario generation within Just Eat Takeaway.com | Max Knobbout | 32m |
Just Eat Takeaway.com estimates causal effects from city-level quasi‑experiments using difference-in-difference and synthetic‑control style matching, augments… | ||
| Practical code archaeology | Judith van Stegeren | 32m |
Claim: about 70% of successful code archaeology comes from talking to people, not new tools. Use concrete techniques: git blame/TIG with a rainbow age color… | ||
| AI for Good- Then and Now | Marijn Markus | 37m |
Models used on Ebola fundraising lead to 10.6 million euros in emergency aid, spatial crime models predict neighborhood stabbing rates, yield-prediction… | ||
| Why does everyone need to develop a machine learning package? | Andrei Alekseev | 28m |
Claims that building an internal ML package yields tested, reusable pipelines and business value backed by concrete work: team authors Aetna, moves from… | ||
| A Tour of the Many DataFrame Frameworks | Harizo Rajaona | 35m |
Compares pandas, Modin, Dask and Polars with concrete internals: API compatibility and migration gaps, partitioning strategies, memory layouts (NumPy arrays vs… | ||
| Turning your Data/AI algorithms into full web apps in no time | Vincent Gosselin | 38m |
Claims that Taipy cuts pilot-to-production cost and time (example: a cash‑flow app delivered for ~€60k with 1.5 people versus a €600k Java project with eight… | ||
| Everything in its Right Place: Optimising Ranking in Online Grocery | Bas Vlaming | 31m |
Mobile grocery ranking combines three concrete pieces: a next-basket repurchase probability model (CARP) trained on delivery histories and product periodicity… | ||
| Closing Notes | Andy Sava | 04m |
Event totals and outcomes are concrete: 45 000 Euros raised and handed to num Focus, two keynotes, six committee members, 11 sponsoring companies, and 21 talks… | ||
| How to create a Devcontainer for your Python project 🐳 | Jeroen Overschie | 36m |
Claim: a reproducible development environment is fully defined and runnable with just two files and a base image. Tools actually used: Docker, VS Code (VS Code… | ||
| Bulk Labelling Techniques | Vincent D Warmerdam | 33m |
Claim: few‑shot embedding selection reaches ~74% accuracy with four seed sentences using KNN while a bag‑of‑words model needs about 500 labeled examples to… | ||
| Thompson sampling for personalising a car brands advertisements | N. van Engelenhoven, J. Hamerlinck | 41m |
Claims a working system that uses Thompson sampling with hierarchical Bayesian priors to choose among 24 creative variants (2 models × 4 concepts × 3… | ||
| Significant Roadblocks to Usefulness for Jupyter Notebooks and a Recipe to Fix them | G. Michaelson | 33m |
Argues Jupyter-style notebooks remain a scratch pad architecture: cells run out of order, hence the restart the kernel button, and cloud collaboration often… | ||
| Becoming a Pokémon Master with DVC: reproducible machine learning experiments | Rob de Wit | 31m |
A working Pokémon sprites classifier gets built to predict water types and returns a sample probability of 0.87 with a small confusion matrix of false… | ||
| Data Storytelling through Visualization | Marysia Winkels | 31m |
Demonstrates that specific design choices — chart type, axis limits, color, title and annotations — change conclusions; uses Anscombe and same stats different… | ||
| Is it a predictive model? Is it causal inference? Well... It is running a greenhouse. | Ruben Mak | 35m |
Automation of greenhouse irrigation at Source combines concrete claims and tools: a multi stage approach first trains a predictive model of water uptake on non… | ||
| Predictive Maintenance | Anjan Prasad Gantapara, Hamideh Rostami | 40m |
Claims production ML models predict maintenance within three months for ASML lithography machines using component-level classifiers and transfer and active… | ||
| How to not pull your hair out while providing data to the business | Lars Hanegraaf | 28m |
Claims that applying unit testing for data pipelines in Python prevents breaking dashboards and models by testing a refactored transform function end to end… | ||
| FuzzyTM: a Python package for fuzzy topic models | Emil Rijcken | 28m |
FuzzyTM is a Python library that trains fuzzy topic models (FSA, FSAW, FSAV, FSAE) and produces topic embeddings for downstream tasks; it implements LSA/SVD… | ||
| Lowering the barrier for ML monitoring | Wesley Boelrijk | 37m |
Practical recipe: expose Prometheus metrics from a FastAPI app and visualize with Grafana, instrumenting Prometheus and Grafana plus Python client counters and… | ||
| nothing matches | ||