Group items tagged

"As machine learning becomes more powerful, the field's researchers increasingly find themselves unable to account for what their algorithms know - or how they know it."

"The current scandal about the mis-use of Facebook data to manipulate elections feels like a pivotal moment in the recent history of the internet and its growing power over societies around the world. "

Google Sheets is getting smarter. After adding the machine learning-powered "Explore" feature last year, which lets you ask natural language questions about your data, it's now expanding this feature to also automatically build charts for you. This means you can now simply ask Sheets to give you a "bar chart for fidget spinner sales" and it will automatically build one for you.

"The ability for machines to see, understand and interact with the world is growing at a tremendous rate and is only increasing with the volume of data that helps them learn and understand even faster. Big data is the fuel that powers AI."

Mentioned at several KIN events including the Lessons Learned RoundTable

Although we'll be writing a lot here about the potential of social media to help people tell their stories, share ideas, start and continue conversations, it is seldom enough on its own. In fact, it is still very much a minority medium in the field of local community action - however powerful it can be, as shown by the work of hyperlocal bloggers (examples here, and we'll be mapping more).

Recommended by Luc Glasbeek at the 2014 Autumn workshop

I wonder what plans Warwick Business School have for expansion into China. I must find out.

Every time I go to WBS, it feels like China!

Use of data analytics on HR data to provide insights to improved employee retention.

McKinsey have developed an approach to retention: to detect previously unobserved behavioural patterns, they combine various data sources with machine-learning algorithms. Workshops and interviews are used to generate ideas and a set of hypotheses. Over time they collected hundreds of data points to test. Then ran different algorithms to get insights at a broad organisational level, to identify specific employee clusters, and to make individual predictions. Finally they held a series of workshops and focus groups to validate the insights from our models and to develop a series of concrete interventions. The insights were surprising and at times counterintuitive. They expected factors such as an individual's performance rating or compensation to be the top predictors of unwanted attrition. But analysis revealed that a lack of mentoring and coaching and of "affiliation" with people who have similar interests were actually top of list. More specifically, "flight risk" across the firm fell by 20 to 40 percent when coaching and mentoring were deemed satisfying.

McKinsey have developed an approach to retention: to detect previously unobserved behavioural patterns, they combine various data sources with machine-learning algorithms. Workshops and interviews are used to generate ideas and a set of hypotheses. Over time they collected hundreds of data points to test. Then ran different algorithms to get insights at a broad organisational level, to identify specific employee clusters, and to make individual predictions. Finally they held a series of workshops and focus groups to validate the insights from our models and to develop a series of concrete interventions. The insights were surprising and at times counterintuitive. They expected factors such as an individual's performance rating or compensation to be the top predictors of unwanted attrition. But analysis revealed that a lack of mentoring and coaching and of "affiliation" with people who have similar interests were actually top of list. More specifically, "flight risk" across the firm fell by 20 to 40 percent when coaching and mentoring were deemed satisfying.