Making pop-ups less annoying and more effective with ML: 1/3: The Concept

Thanks for all the recent signups to the group. We’re fortunate and humbled to have so many smart and skilled people here.

This group was founded on the promise of sharing the takeaways from the development of machine learning based solutions to real world problems.

It’s time we do something about the promise!

PROBLEM

Although various website pop-ups (modals) aren't one of the Grand Challenges of the 21st century, they're certainly somewhat of a first world problem. Over the years, they’ve earned a reputation of being extremely annoying and intrusive. Yet businesses don’t want to drop them, because they often work.

The questions we asked ourselves were: is there a way to make pop-ups less annoying and more effective? Can we find a way to display the popup in the right place at the right time to maximize the conversion probability and minimize the churn probability?

Today, we’ve already one month into the development of machine learning based solution to this problem.

SERIES AGENDA

We plan to host three meet ups on this topic.

Meetup # 1 (this one):

We’ll explain the concept and its technical overview and machine learning architecture as well as data science challenges. We'll explain the technology stack, which (as of this moment), includes the following technologies:

• Database: MongoDB - powerful no-SQL database, great for quick development of the application and extremely easy to perform pivots which are crucial in the initial phase of creating the product

• Language: Python 3.6: extremely powerful in building light web applications with great integration with ML libraries

• ML library: TensorFlow: industry standard with huge community, great integration with Python

• Web framework: Tornado: Simple yet powerful tool for building scalable web application and developing REST API

• Virtualizing environment: Docker: industry standard container technology. Adds important layer of abstraction of operating system and enhances process of application deployment and portability.

Meetup # 2 (July):

We plan to show a fully functional application.

Meetup # 3 (August):

At the third meetup, we’ll share our lessons in scaling the application.

NETWORKING

Let's have a beer or two after the event.