PyData London 2018
2018-04-26..2018-04-29
London, United Kingdom
[PyData London 2016, PyData London 2017, PyData London 2018, PyData London 2019, PyData London 2022, PyData London 2024, PyData London 2025, PyData London 2026]
| title | author | length |
|---|---|---|
| Democratising data journalism | Charles Boutaud | 40m |
Bureau Local builds a collaborative newsroom that turns national open datasets into regional investigations using Python, publishing 169 local stories and 40+… | ||
| Keynote: Making the Big Data ecosystem work together with Python | Holden Karau | 41m |
Shows concrete fixes: JVM↔Python double-serialization via Py4J and custom pickling causes huge overhead and copying; replacing that path with Apache Arrow and… | ||
| Keynote: Learning programming and science with Scientific Python | Emmanuelle Gouillart | 52m |
Claims that scikit-image pairs cleanly with scikit-learn for real science get demonstrated with concrete pipelines: Daisy features into a random forest for… | ||
| Keynote: Artistic Applications of Artificial Intelligence | Luba Elliott | 39m |
Survey of concrete outcomes not slogans artists use DeepDream outputs and variational autoencoder reconstructions to remake Blade Runner and receive a… | ||
| Automatic Tagging of Short Texts with scikit-learn and NLTK | Gilbert François Duivesteijn | 01h 09m |
A practical walkthrough of building an automatic tagger for short texts (Slack messages, email subjects, tweets) using scikit-learn and NLTK. It explains the… | ||
| Computer Vision: An (Un?)Expected Journey, with Keras and Tensorflow | Rodolfo Bonnin | 01h 29m |
Hands-on work uses Keras and TensorFlow notebooks (Colab-ready) with concrete pipelines: train classifiers on Fashion-MNIST (simple dense nets hit ~70% then… | ||
| Test-Driven Data Analysis | Nick Radcliffe | 01h 29m |
Shows a concrete workflow: install tdda with pip, use reference tests to capture inputs and outputs and auto-update golden files, and run tdda… | ||
| Apache Airflow in the Cloud: Programmatically orchestrating workloads w/ Py | Satyasheel, Kaxil Naik | 01h 25m |
Practical tutorial that shows Airflow running real pipelines: code examples create dynamic DAG from CSV, use a GCS-to-BigQuery operator to load Avro files and… | ||
| Deep Probabilistic Methods with PyTorch | Chris Ormandy | 01h 22m |
Starts from Bayesian linear regression and implements VI end to end in PyTorch 0.4, with notebooks on GitHub and Adam used for optimization. Derivations… | ||
| Anomaly Detection | Nick Radcliffe | 01h 16m |
Workshop previews a concrete anomaly workflow: use Python, pandas and the TDDa library to write concrete constraints (uniqueness, min/max, allowed values), run… | ||
| Making computation easier with cool Numpy tricks | Kirit Thadaka | 01h 27m |
Concrete claims and demos: Numpy yields real speedups (range/sum benchmarks show Python lists several times slower—examples ~6x and ~36x), and some routines… | ||
| Network Science, Game of Thrones and US Airports | Mridul Seth | 01h 27m |
PyData London 2018
Network Science deals with modeling and analyzing networks. Politics, mathematics, law, computer science, finance data can be modeled as… | ||
| Understanding and diagnosing your machine-learning models | Gael Varoquaux | 01h 33m |
Claims concrete diagnostics not platitudes: uses scikit-learn metrics (explained variance, mean absolute error), cross_val_score, DummyClassifier and… | ||
| Hypothesis Testing with SciPy | Hillary Green-Lerman | 01h 11m |
Gives concrete recipes and outcomes: uses SciPy ttest_1samp and ttest_ind on toy and product data, shows when p values cross 0.05 and why stopping early biases… | ||
| Follow the gradient: an introduction to mathematical optimisation | Gianluca Campanella | 01h 28m |
Workshop explains which optimization problems are easy, which are hard, and shows working code: linear and convex quadratic formulations solved with CVXpy… | ||
| Sentence embeddings for automated factchecking | Lev Konstantinovskiy | 38m |
Uses InferSent sentence embeddings trained on SNLI with GloVe inputs and a BiLSTM max‑pool encoder to spot verifiable claims in political subtitles. Annotated… | ||
| CatBoost - the new generation of gradient boosting | Anna Veronika Dorogush | 38m |
Claim: CatBoost outperforms other GBM libraries on public benchmarks and ships with concrete implementations: symmetric trees for very fast prediction… | ||
| Data, Science: Investigating 1 Million Galaxies with Humans and TensorFlow | Mike Walmsley | 39m |
Combines crowdsourced labels and TensorFlow to analyze roughly 1 million galaxies using volunteers treated as classifiers with per-user confusion matrices… | ||
| Python Doesn’t Have to Be Slow: Speeding Up a Large-Scale Optimization Algorithm | Dat Nguyen | 31m |
A Peer‑to‑Peer portfolio allocation system uses a greedy heuristic and tight engineering to turn an intractable integer program into a fast practical pipeline… | ||
| Why giving your algorithm ALL THE FEATURES does not always work | Thomas Huijskens | 38m |
Claim: throwing every variable at a model often worsens performance because of collinearity and weak signals with scarce samples. Shows concrete methods… | ||
| Touchdown Localisation with aircraft flight data | Jonathan G. Pelham | 35m |
Parses QAR/FDR binary into CSV, uses Python (pandas, GeoPandas, folium, overpy, shapely) to align radio-altimeter profiles with runway geometry and produce… | ||
| Beyond word2vec: GloVe, fastText, StarSpace | Konstantinos Perifanos | 39m |
Compares word2vec, GloVe, fastText, StarSpace by claim, tool and result: uses summed context vectors and off‑the‑shelf embeddings as ML features, reaches about… | ||
| Creating correct and capable classifiers | Ian Ozsvald | 42m |
Framework gives concrete checks: start with a dummy classifier baseline (Titanic example ≈0.62), then an untuned random forest reliably improves scores (≈0.83)… | ||
| Demystifying pandas internals | Marc Garcia | 42m |
PyData London 2018
pandas high-level API makes it easy for newcomers to do data wrangling and analysis, without having to know much about data structures or… | ||
| Evaluating fairness in machine learning with PyMC3 | Oliver Laslett | 40m |
Claims a concrete probabilistic threshold model for stop‑and‑search, implemented first in PyMC3 then PyStan after hitting a modeling limitation, and fits UK… | ||
| A Data Science Approach to Systemic Risk | Nikolai Nowaczyk | 36m |
Simulated financial systems are generated with random graph generation and fed into an open-source C++ risk engine connected via a Python stack; the pipeline… | ||
| Bayesian forgetting and Random Effects for forecasting TV ratings | Ruadhán Stokes | 37m |
Claim: Bayesian filtering with a program-level random effects model and an estimated forgetting factor improves Channel 4 TVR forecasts versus fixed-forgetting… | ||
| Emphasising Relationships in the BBC's Data Using Technologies of the Semantic Web | Theo Windebank | 39m |
Claim: using RDF and ontologies actually links BBC data silos into a usable knowledge layer. Tools shown: SPARQL queries against a dbpedia endpoint using… | ||
| More About Generators | James Powell | 41m |
Generations of Python features start with PEP-255 and PEP-342 and the claim here is concrete: generators act as coroutines and sequencing primitives but rarely… | ||
| Winning with Simple, even Linear, Models | Vincent D Warmerdam | 44m |
Claims that linear models beat complex ones when you apply concrete tricks: add an interaction to solve XOR, use radial basis seasonal features and exponential… | ||
| A/B Testing with Style | Alistair Lynn | 41m |
Walks through a working A/B system called Jacquard and why it replaces per-user DB rows with 300 buckets hashed from user IDs, so tests need no signup writes… | ||
| Python at Massive Scale | Stephen Simmons, Neil Slinger | 45m |
Athena runs a mono repo of 35m LOC/150k modules served from a custom Hydra object DB with import hooks and continuous deployment, supporting about 1,600 active… | ||
| Auto-encoders in the wild... of telco land | Guillermo Christen | 31m |
Implements MXNet symbolic autoencoders with a 200-dimensional embedding and trains with cosine loss + transpose reconstruction while adding a softmax task to… | ||
| Multi-touch Attribution: What am I training for? | Sri Sri Perangur | 37m |
Compares Markov, Shapley and rule baselines inside a simulated marketing environment built from Skyscanner traces, trains and stress-tests methods, and pushes… | ||
| RNN sequence labeling for document parsing in Tensorflow | Carsten van Weelden | 39m |
CV parsing uses Tensorflow to train a BLSTM sequence labeling model that cuts errors by about 30% error reduction compared with CRF/HMM baselines… | ||
| JupyterHub from the Ground Up with Kubernetes | Camilla Montonen | 37m |
Explains how to deploy a secure, production-ready JupyterHub on Kubernetes using the official Helm chart, shows the hub + configurable HTTP proxy + per-user… | ||
| Searching for Shady Patterns: Shining a light on UK corporate ownership | Adam Hill | 40m |
Built a graph database mapping the UK beneficial-ownership register (about 4.5 million companies and 4 million people), ingesting Companies House dumps plus… | ||
| Unsupervised Anomaly Detection with Isolation Forest | Elena Sharova | 33m |
Isolation Forest is an ensemble of extremely randomized trees implemented in scikit-learn and the claim is concrete: on KDD Cup 99 and SF variants it reaches… | ||
| Databases for Data Science | Alexander CS Hendorf | 45m |
Shows concrete claims and runnable tools not buzzwords: compares Postgres, MongoDB, Cassandra and Neo4j with query examples, mentions MongoDB aggregation… | ||
| Data Deduplication using Locality Sensitive Hashing | Matti Lyra | 40m |
Explains that minhash + locality sensitive hashing turns Jaccard set similarity into a probabilistic collision test and shows an open source Python… | ||
| Reliably forecasting time-series in real-time | Charles Masson | 41m |
Claims an algorithm that mixes seasonal SARIMA/Prophet-style fits with a non‑seasonal strategy that finds linear sub-windows and takes a weighted median of… | ||
| Be good (and don't be evil): how to audit your work for fairness and inclusion | Aileen Nielsen | 42m |
Walks through concrete audits: case studies (Latanya Sweeney ads, COMPAS risk scores), legal pressure in the EU and US, and hands‑on fixes that actually change… | ||
| Visualising NLP pipelines with Pynorama | Slavi Marinov | 34m |
A Python visualization system called Pynorama (Panorama) visualizes whole NLP corpora, metadata and model pipelines, not just one chart. It ships as a Flask… | ||
| Do tyres dream of electric clouds? | Tom Alisi | 40m |
Connesso mounts sensors in the tyre tread groove and streams pressure temperature angular velocity and patch angle via a dongle to a cloud pipeline that runs… | ||
| How will the new EU data protection laws affect your work? | Will Hardy | 40m |
Regulation widens scope and raises penalties — expect fines of 4% of global revenue and uniform rules across the EU — and specific operational demands: allow… | ||
| Using Survival Analysis to understand customer retention | Lorna Brightmore | 35m |
Uses Kaplan–Meier and Cox models to turn partial subscriptions into concrete LTV inputs, running with lifelines and scikit-survival, encoding high‑cardinality… | ||
| Big Data Oceanography | James Munroe | 38m |
Claims that oceanography workflows scale from single workstations to cluster-backed processing using xarray and dask, handling tens to hundreds of terabytes of… | ||
| Who's singing? Automatic bird sound recognition with machine learning | Dan Stowell | 40m |
System learns species from 10‑second mobile recordings, sending audio to a cloud service that returns a ranked list and stores data as open archives. Feature… | ||
| Planes, Trains, and Skateboard Shoes - Bayesian methods in engineering and product design | Jim Parr | 41m |
Uses Bayesian optimization with Gaussian processes and an acquisition function to tune real designs: a ducted‑fan with eight geometry parameters evaluated by… | ||
| PyData London 2018 Lightning Talks - Day 1 | 53m | |
PyData lightning compiles short results and concrete tools: NumFOCUS fundraising and community impact, a worldwide pandas sprint that fixes 300 pandas docs… | ||
| PyData London 2018 Lightning Talks - Day 2 | NumFOCUS representative | 45m |
Day‑long lightning rounds deliver concrete wins: a pandas sprint yields 2–3 pull requests ready for review fixing string and docs issues; an astronomy project… | ||
| nothing matches | ||