Meetup #3: Up to Data

Hello Data & AI London!

Thank you so much for joining us as we kicked off '22 with some great talks!

We've had an absolutely fantastic turnout to both events so far and look forward to continuing that as we head into the new year.

We are super excited to welcome you all to our 3rd meet up & the 1st of the '23 Data & AI events calendar. We're getting started with some of the hottest trends in Data for '23 (see speakers below) once again at Mesh-AI HQ in Liverpool Street.

Doors will be open from 6pm talks will start around 6.30pm. We will have food and both alcoholic and non alcoholic beverages provided throughout the evening along with the following line up:

Talk #1 : Andrew Jones | GoCardless | Driving data culture change with Data Contracts

Synopsis:

At GoCardless we’ve been implementing Data Contracts since 2021, using it as our vessel to increase the quality of data and drive a culture change in order to become a true data-driven organisation - one that really values its data and uses it to drive our products and business.

Following this talk you’ll understand the problems we’re trying to solve with Data Contracts, and how through both the tooling *and* by working with people across the organisation we’re changing our data culture to one where we’re much more deliberate about the data we produce, manage, and consume, leading to better data-driven outcomes across the business.

Talk #2 : Steve Goodman | Tide | ML Explainability - From correlation to causation

Synopsis:

Most of the time, we build machine learning models first and foremost to make predictions. And as long as those predictions prove to be accurate in an operational setting, and some basic assumptions are met (such as - future patterns will behave like historic patterns), then we are good. We know that correlation is not causation, but that correlations can still be useful to make accurate predictions.

But when it comes to interpreting those models in order to understand which factors are driving those predictions - either for regulatory purposes, or from a commercial policy perspective, then we need to care more about whether the drivers can be considered causal or not.

In Steve’s work he gets asked a lot of questions along the lines of “What IF”; We had done something different? What would the outcome have been? What levers can we pull going forward to improve business performance?

Unless we attempt to understand cause and effect then it's difficult to be confident about the recommendations we’re making.

After briefly introducing the background to causal inference theory Steve will walk through a quick practical example (from a dataset in the public domain) where using conventional ML models and explainability methods can lead to incorrect inferences, and some remedies using causal analysis to help uncover the real drivers of user behaviour.