Decision Mechanics

Insight. Applied.

The top three spreadsheet errors of the decade

By

IBM just published an article describing what they claim to be the top three spreadsheet errors of this decade—so far. Spreadsheets errors caused

  • the sale price of Tibco Software to be overstated by $100m
  • fatal errors in an oft-quoted fiscal austerity research project
  • 10,000 sports fans to miss a synchronised swimming event at the 2012 Olympics

Some companies I work with have started taking real steps to get spreadsheet anarchy under control. But, there are still way too many organizations who have yet to even understand they have a spreadsheet addiction. I even know of organizations who are still authorizing the development of new business critical Excel applications within departments lacking any IT expertise.

Insanity.

As enterprise software challenges go, this isn’t a particularly difficult one to fix—technically. It’s more of a people problem. First, many managers don’t understand that having key business processes run via locally maintained spreadsheets is a huge risk. And, when they do, they find spreadsheet owners unwilling to relinquish control.

Like any addiction, admitting to the problem is the first step on the road to recovery.

Spark 1.5.0 released

By

Spark 1.5.0 has now been released—and it’s a significant one for the data science community. Databricks, in their announcement blog post, state

Another major theme of this release is data science: Spark 1.5 ships several new machine learning algorithms and utilities, and extends Spark’s new R API.

Improvements of note include better coverage for the pipeline API and an MLlib API for SparkR.

Apple steps up recruitment of machine learning experts

By

Apple is looking to recruit another 86 artificial intelligence experts, according to an article on VentureBeat.

The recruitment drive is due to concerns that they are falling behind Google, Amazon, Facebook and Microsoft in the area of machine learning. Competitors seem to be stealing a march on Apple by developing services that can anticipate users’ requirements—services that rely on sophisticated machine learning capabilities.

One of the challenges facing Apple is their privacy policy. Machine learning is data hungry and researchers crave lots of high quality data. As Apple leaves a substantial proportion of data on users’ devices—rather than storing it in the cloud—they have less data to play with than their peers. Less data means less effective machine learning solutions and, critically, also makes Apple a less attractive employer for ambitious PhDs.