Jeremie Charlet | Devoxx

Devoxx UK 2019
from Wednesday 8 May to Friday 10 May 2019.

Entrepreneur, polyglot developer, with a thrill for learning and passionate about ML/DL.

CTO at Trackener, a tech startup on horse care, in charge of the cloud platform previously backend developer at TNA Gov UK and Worldline by Atos

Mentor, mentee and co-lead of Machine Learning Study group in meet-a-mentor community.

See also www.trackener.com

Do we know our data as good as we know our tools?

Conference


For many of us who are developer turning data scientist, we are always concerned about how to build a model, train it, etc... And yes, we want the best accuracy (close to 99%).

But as every seasoned data scientist will always advise us, we need first and foremost to understand our data, ensure it’s clean and prepared before doing any training on it.

During the conference, we will explore multiple problems occurring during data analysis or preparation and for each a technique to solve them (from a list of them). You will go away with a number of resources to explore at your own pace.

We will cover these categories of problems:

  • dirty data
  • disparate datasets - needing normalisation
  • too much information to process
  • and others…

We will cover some of these techniques:

  • analysis - detecting misleading data, outliers, specific time series issues
  • cleaning - deal with missing/ambiguous values, outliers, generating synthetic data, resampling
  • preparation - using statistical and physics functions, dimensionality reduction, feature selection, resampling

And using different kinds of plots relevant at different stages.

Sign-in
Make sure to download the Android or iOS mobile schedule.