AI News, Machine Learning & Artificial Intelligence Archives artificial intelligence

150 successful machine learning models: 6 lessons learned atBooking.com

150 successful machine learning models: 6 lessons learned at Booking.com Bernadi et al., KDD’19 Here’s a paper that will reward careful study for many organisations.

We found that driving true business impact is amazingly hard, plus it is difficult to isolate and understand the connection between efforts on modeling and the observed impact…

Our main conclusion is that an iterative, hypothesis driven process, integrated with other disciplines was fundamental to build 150 successful products enabled by machine learning.

On the contrary, developing an organisational capability to design, build, and deploy successful machine learned models in user-facing contexts is, in my opinion, as fundamental to an organisation’s competitiveness as all the other characteristics of high-performing organisations highlighted in the State of DevOps reports.

You’ve probably heard of Booking.com, ‘the world’s largest online travel agent.’ Delivering a great experience to their users is made challenging by a number of factors: With around 150 models now in production, you won’t be surprised to hear that machine learning has touched many parts of the Booking.com experience.

The Problem Construction Process takes as input a business case or concept and outputs a well-defined modeling problem (usually a supervised machine learning problem), such that a good solution effectively models the given business case or concept.

Booking.com go to some lengths to minimise the latency introduced by models, including horizontally scaled distributed copies of models, a in-house developed custom linear prediction engine, favouring models with fewer parameters, batching requests, and pre-computation and/or caching.

The large majority of the successful use cases of machine learning studied in this work have been enabled by sophisticated experiment designs, either to guide the development process or in order to detect their impact.

The last word: Hypothesis driven iteration and interdisciplinary integration are the core of our approach to deliver value with machine learning, and we wish this work to serve as guidance to other machine learning practitioners and spark further investigations on the topic.

Customer churn classification using predictive machine learning models

Customer attrition, customer turnover, or customer defection — they all refer to the loss of clients or customers, ie, churn.

In this article, we will explore 8 predictive analytic models to assess customers’ propensity or risk to churn.

These models can generate a list of customers who are most vulnerable to churn, so that business can work towards retaining them.

This is critical to business because it is often more expensive to acquire new customers than to keep existing ones.

The features in this dataset include the following:· demographic data: Gender, SeniorCitizen, Partner, Dependents· subscribed services: PhoneService, MultipleLine, InternetService, OnlineSecurity, OnlineBackup, DeviceProtection, TechSupport, StreamingTV, StreamingMovies· customer account information: CustomerID, Contract, PaperlessBilling, PaymentMethod, MonthlyCharges, TotalCharges, Tenure Target is Churn, which has binary classes 1 and 0.

For example, gender (whether male or female) and phone related services, customers are equally likely to churn, because the ratio of churn and non-churn customers are the same.

For example, those with month-to-month contract are more likely to leave, which is logical as they are not bounded by yearly contracts.

So, I have carefully selected these separation boundaries to attempt to separate the churn and non-churn cases.

It means the model has performed better than random guesses of 0.5, the diagonal line.

And business can also focus on this list of features to understand whether a customer will likely to churn or not.

'Session 15- More human than human: artificial intelligence in the archives'

'Session 15- More human than human: artificial intelligence in the archives' by Dr Gregory Rolan, Glen Humphries, Lisa Jeffrey and Tatiana Antsoupova (ASA ...

Machine learning meets NASA archives

To celebrate 60 years of NASA, the Google Arts & Culture Lab explored NASA's archive of 127000 historic images stretching back to 1915, and analyzed them ...

Artificial Intelligence and Machine Learning Solution

Learn how AI/ML can help you make better decisions. For more information about Cisco AI/ML computing solutions, please visit

George Hotz | Programming | Context Tree Weighting: compression = AI | Part1

Date of stream 26 Nov 2018. Live-stream chat added as Subtitles/CC - English (Twitch Chat). Stream title: Context Tree Weighting: compression = AI | Part1 ...

When Machine Learning Becomes AI

Even among experts, the terms “machine learning” and “AI” can mean very different things. When does it truly become artificial intelligence? a16z general ...

Webinar - How Artificial Intelligence brings value to your existing archives using Elvis DAM

AI-powered Elvis DAM unveils hidden content treasures and reduces production costs As a publisher, you're constantly challenged to create compelling content ...

Using machine learning and AI to reduce hospital readmissions

The introduction of cloud computing and machine learning is helping researchers identify ways to reduce hospital readmissions for chronic conditions and ...