Vorige Meetup

PyData Amsterdam: a new correlation coefficient Phi_k and a Dash demo and more

Deze Meetup is voorbij

103 personen zijn geweest

KPMG Hoofdkantoor

Laan van Langerhuize 1 · Amstelveen

Hoe vind je ons

Public transport: You can take bus 356 from Bijlmer arena, this bus goes every 6 minutes Metro line 51 stops near the office (oudekerkelaan) and then it is just a 5 minute walk. Please take the entrance on the right :)

Locatieafbeelding van evenementslocatie

Wat we doen

Hi all,

Time for the first meetup of 2019!

# A word to our sponsor!

We want to thank the kind folks of KPMG for hosting the meetup and providing for good food and drinks!

Thanks KPMG!

Want to work at KPMG: https://home.kpmg.com/nl/nl/home/werken-bij-kpmg.html

# Program

Coming to our meetup now: we have three amazing talks. We will start with food around 18:00. At 19:00 the first talk will start, the second one at 19:30 and the third one at 20:15.

There will be enough time to mingle before, in-between, and after the talks.

Around 21:30 the bar will close!!

One final remark: KPMG will print name badges for the attendees, so make sure you are on the list!

## Phi_k correlation

The calculation of correlations between paired data variables is a standard tool of analysis for every data analyst. This presentation will be about a new and practical correlation coefficient, phi_K, which works consistently between categorical, ordinal and interval variables. It is based on several refinements to Pearson's hypothesis test of independence of two variables and captures non-linear dependency.

Emphasis is paid to the proper evaluation of statistical significance of correlations and to the interpretation of variable relationships in a contingency table, in particular in case of low statistics samples. Two practical applications are discussed. The presented algorithms are easy to use and available through a public Python library.


pip install phik


Rose Koopman is a data scientist at KPMG, working on data driven solutions for clients in different domains. Before joining KPMG she did a PhD in high energy physics at CERN.

## Dash

Dash in a python framework utilized to create interactive web applications for doing and showing analyses. It is a powerful tool which will allow users to create anything from a single interactive plot to a full blown dashboard.
This talk aims to show you how to get started using dash in your daily work, making a template data exploration tool that can aid in exploring new data or showing results to clients.
In addition we will demo our dash-app that uses phi_k!

Susanne is a Data Scientist working at KPMG in all different sectors, having developed an interest in NLP, Deep learning and visualizations in the recent months. Before joining KPMG she worked in IT and completed a masters in Medical Physics.

## Third talk

*Postponing work in an optimal way*

Performance evaluation and optimal control of processes are crucial challenges for any business. One common approach would be to use Machine Learning (ML) on a past data to uncover important patterns and only afterwards model this within an Operations Research (OR) framework. In this talk, I will present a novel method to combine ML and OR. Although, the technique is highly generic, in this session we will concentrate only on one process - “postponing work in an optimal way”. After discussing the pros and cons of both traditional OR and ML, we will see how one can benefit from their synergy.

### Bio:

Asparuh Hristov, data scientist at bol.com


That's it, we hope to see you the 10th!

The PyData Amsterdam committee

Deelnemers (103)