PyData Berlin 2016
2016-05-20..2016-05-21
Berlin, Germany
[PyData Berlin 2014, PyData Berlin 2015, PyData Berlin 2016, PyData Berlin 2017, PyData Berlin 2018]
| title | author | length |
|---|---|---|
| Using small data in the client instead of big data in the cloud | Anton Dubrau | 37m |
Claim: ship complete Transit App schedules to the device and query them client-side so UI is instant and offline, avoiding round trips and server load… | ||
| Predictive modelling with Python | Olivier Grisel | 58m |
PyData Berlin 2016
www.pydata.org
PyData is a gathering of users and developers of data analysis tools in Python. The goals are to provide Python enthusiasts… | ||
| Frontera: open source, large scale web crawling framework | Alexander Sibiryakov | 43m |
A scalable crawler framework is built and released as Frontera, using Scrapy, Kafka and HBase with per‑host queues, local state cache and DNSmasq; the team… | ||
| Functional Programming in Python | Daniel Kirsch | 39m |
Claims that Python supports functional style enough to be useful, demonstrates partial function application, currying, closures, decorators and function… | ||
| One in a billion: finding matching images in very large corpora | Ryan Henderson | 15m |
An open-source reverse-image-search system called image-match claims a complete free solution that encodes perceptual hashes, uses Python and numpy, and… | ||
| Dealing with TBytes of Data in Realtime | Nils Magnus | 45m |
Claim: stitch together hundreds to thousands of single SQL instances into a shared‑nothing cluster and get near‑real production scale. Uses Crate on an unused… | ||
| Classifying Search Queries Without User Click Data | Abhishek Thakur | 34m |
Claims a pipeline that classifies queries using LSTM and convolutional nets plus XGBoost and word2vec features derived from SER titles and Wikipedia… | ||
| BigchainDB : a Scalable Blockchain Database, in Python | Trent McConaghy | 43m |
BigchainDB presents a decentralized, permissioned database built in Python on top of rethink DB, claiming sustained throughput of 1 million writes per second… | ||
| How to Trick a Neural Network | Julia Evans | 27m |
Implements the adversarial-examples recipe from "Explaining and Harnessing Adversarial Examples" using GoogleNet downloaded into Caffe and runnable as a… | ||
| Machine Learning at Scale | Nathan Epstein | 27m |
scikit-learn falls short on very large datasets so the stack uses PySpark and MLlib on AWS with HDFS (optionally managed via Flint Rock) to run distributed… | ||
| What every data scientist should know about data anonymization | Katharina Rasch | 36m |
Numerical claims and concrete failures decide whether to watch: Netflix prize example shows ratings reidentification from two ratings plus dates; mobile-phone… | ||
| Accelerating Python Analytics by In-Database Processing | Edouard Fouché | 37m |
Claims that pushing Python operations into SQL speeds analytics by keeping data in-place and using DB features. The demo builds an ibmdbpy prototype that… | ||
| Python and TouchDesigner for Interactive Experiments | Jessica Palmer | 27m |
Uses Python inside TouchDesigner to run a 60 Hz node network that takes live signals from Muse EEG headsets, replicates player pipelines to any N (prototype… | ||
| Let's play Space Invaders! | Maciej Jaskowski | 35m |
Reimplementation of Deep Q-Network (DQN) trains on raw grayscale frames (stacked 4×80×80) from the Arcade Learning Environment, learns actions… | ||
| Python Data Ecosystem: Thoughts on Building for the Future | Wes McKinney | 58m |
Discussion centers on concrete work not platitudes: a C++11 library and bindings that implement a columnar in‑memory format called Arrow, a tiny file format… | ||
| What's new in Deep Learning? | Kashif Rasul | 37m |
Survey of concrete advances and how to use them: Glorot/Xavier and He initialization for stable signal flow, BatchNorm applied per-mini‑batch with running… | ||
| Designing spaCy: Industrial-strength NLP | Matthew Honnibal | 40m |
spaCy is built as a C extension with a Python API and implemented in Cython from the start, enabling GIL-free multithreading and a compact memory‑managed core… | ||
| Data Integration in the World of Microservices | Valentine Gogichashvili | 37m |
Zalando explains why isolated service databases break ETL and how PostgreSQL’s logical replication actually streams row‑level changes out of separate AWS… | ||
| Bridging the gap: from Data Science to service | Daniel Moisset | 37m |
Data science usually accounts for 20–30% of delivery effort while the rest is engineering, ops and coordination; projects succeed when teams verify raw data… | ||
| The "Kwargh!" Problem | James Powell | 32m |
Claim is that Python is essentially impossible to fully statically analyze but useful heuristics can recover many forwarded keyword args. Uses inspect and ast… | ||
| Usable A/B testing – A Bayesian approach | Nora Neumann | 38m |
Bayesian A/B testing uses conjugate posterior beta distributions and Monte Carlo sampling (numpy or PyMC) to report probabilities and credible intervals… | ||
| Introduction to Julia for Python Developers | David Higgins | 34m |
A compact overview that claims Julia runs at C-like speeds by JIT compiling through LLVM, uses multiple dispatch for type-specialized performance, and can call… | ||
| Robot uses toddler-like self exploration for the development of body representations | Idai Guertel | 17m |
Robot explores with Nao arm babbling, collects joint angles, target positions and electrical currents, records about 10,000 samples, and trains a feed‑forward… | ||
| pypet: A Python Toolkit for Simulations and Numerical Experiments | Robert Meyer | 25m |
A Python toolkit that actually ties parameter sets to results and writes them into HDF5 storage via a trajectory container, so parameters, NumPy arrays and… | ||
| PySpark in Practice | Ronert Obst, Dat Tran | 38m |
Shows concrete PySpark choices and outcomes: configuration recipes for a 6‑node 64 GB cluster with 1.5× executor overcommit and per‑executor 512 MB Python… | ||
| Zero-Administration Data Pipelines using AWS Simple Workflow | Anne Matthies | 38m |
Floto is an open source Python library (github.com/babbel/floto) that uses AWS Simple Workflow to run scalable pipelines with zero-administration and no master… | ||
| Visualizing research data: Challenges of combining different datasources | Juha Suomalainen | 35m |
A concrete pipeline story: Python scripts ingest Excel, CSV and JSON APIs, read ESRI shapefiles with pyshp, normalize ISO codes and encodings, simplify… | ||
| Spotting trends and tailoring recommendations: PySpark on Big Data in fashion | Martina Pugliese | 35m |
Mallzee builds a per-user recommender that runs on a PySpark cluster, featurizing products with scraped metadata, keywords and 14 hand-tagged brand style… | ||
| Smart Banking - Real Time Driven | Arnab Dutta, Christian Rebernik | 33m |
Number26 describes a concrete pipeline: real time categorisation of transactions using AWS Kinesis, DynamoDB and Amazon Machine Learning plus NLP (stanford… | ||
| Visualizing FragDenStaat.de | Andrej Warkentin | 26m |
FragDenStaat.de data gets scraped from the FragDenStaat.de API into pandas dataframes and plotted with pandas and matplotlib; the dataset contains over 12000… | ||
| Plumbing in Python: Pipelines for Data Science Applications | Thomas Reineking | 36m |
Blue Yonder builds and runs a Python data‑flow library called llamar that declares pipelines (splitter/reducer/fork) separately from execution, uses dill to… | ||
| Bayesian Optimization and it's application to Neural Networks | Moritz Neeb | 34m |
Claims that Bayesian optimization finds better hyperparameters for neural nets with far fewer costly trainings than grid or random search, modeling the outer… | ||
| Python based predictive analytics with GraphLab Create | Danny Bickson | 30m |
Shows how to build and deploy a real predictive app that combines tables, graphs, text and images using GraphLab Create and its SFrame engine, feeding models… | ||
| A full Machine learning pipeline in Scikit-learn vs in scala-Spark: pros and cons | Jose Quesada | 39m |
Compares a full pipeline run twice on the same messy dataset — exploration, cleaning, feature engineering, model fitting — showing that Spark.ml uses… | ||
| Connecting Keywords to Knowledge Base Using Search Keywords and Wikidata | Fang Xu | 31m |
PyData Berlin 2016
The development of large-scale Knowledge Base (KB) has drawn lots of attentions and efforts from both academy and industries recently . In… | ||
| Predicting political views from text | Felix Biessmann | 37m |
Claim: basic machine learning and NLP applied to parliament speeches and party manifestos can predict political affiliation with concrete numbers: about 60%… | ||
| Estimating stock price correlations using Wikipedia | Delia Rusu | 36m |
Claims that Wikipedia text and links can estimate future stock correlations for DAX companies using concrete tools and numbers: article vectors via… | ||
| Brand recognition in real-life photos | Lukasz Czarnecki | 25m |
A brand‑recognition pipeline uses pretrained VGG as a feature extractor and SVMs on Instagram crops to detect six brand pairs (Nike vs Adidas, Starbucks vs… | ||
| Building a polyglot Data Science Platform on Big Data systems | Frank Kaufer | 36m |
Slides: https://speakerdeck.com/fkaufer/polyglot-data-science-platforms-on-big-data-systems 00:00 Welcome!
00:10 Help us add time stamps or captions to this… | ||
| Removing Soft Shadows with Hard Data | Maciej Gryka | 25m |
Claims a pipeline that learns soft-shadow mattes from photorealistic rendering and uses them to remove or edit shadows in RGB photos with only a brush mask. It… | ||
| ExpAn - A Python library for advanced statistical analysis of A/B tests | Jie Bao | 28m |
ExpAn is a Python library that defines a standard input format, standalone analysis functions and a generic result schema and ships on PyPI and GitHub; it… | ||
| Setting up predictive analytics services with Palladium | Andreas Lattner | 39m |
Description
We will introduce Palladium, an open source framework for setting up predictive analytics services. It supports tasks like fitting, evaluating… | ||
| nothing matches | ||