Diagnosing Machine Learning Models

Date: 
Thursday, June 21, 2018 - 19:00
Source: 
Central London Data Science Project Nights
Attendees: 
76
City: 
London

Every problem has their own measure of success and failure. Performance benchmarking gives us the ability to not only compare how a model performs in relation to a specific end goal but also how it fairs in comparison to previous iterations of ideas.

Identifying and evaluating performance metrics is an essential stage of a model’s life cycle. The ability to give a performance diagnosis can give us a greater insight in how to fine-tune our model which in turn will lead to an optimized product.

So what are you waiting for? Sign up to our next Meetup and we’ll show you how to become a machine learning doctor in no time!

Meetup structure:
* Lightning talk on model diagnosis
* Hands on coding
* Pizza and drinks!

Learning outcomes:
We aim to get everyone up to speed in how to pick and optimize performance metrics based on specific machine learning problems.

Software requirements:
We will be using Kaggle kernels, so make sure you sign up for a Kaggle account before the Meetup starts.

Saint Mary's Campus, Imperial College London

St Mary's Hospital, Praed St, Paddington