Decision Mechanics

Insight. Applied.

BBC News using ggplot2 for production-ready charts

By

Chart montage

BBC News are using the R ggplot2 package to create production-ready charts—which illustrates the flexibility and power of this package.

What’s particularly interesting is the approach they’ve taken to training their staff to use R. Initiatives include

  • sharing of tricks and tips through a living cookbook (knowledge base)

  • a six week course comprising six short introductions with links to online resources

  • a dedicated Slack channel for course participants to discuss issues or ask for help

  • an end-of-course project/challenge that exercises the skills developed over the six weeks

The strategies of spreading training over an extended period and embedding it in the workplace have been shown to increase retention and assist in moving from learning to delivering value to the business.

Ranges are for cattle

By

I’m always surprised by how much focus is placed on advanced data science techniques when basic statistical concepts are ignored.

Take fixed point estimates. It’s clear that many of the numbers bandied around in organisations have significant levels of uncertainty associated with them. How many of us can hit the cost of a project, or its completion date, on the nail?

However, there’s little appetite for working with probability distributions. Even offering up a range of possible outcomes results in demands for a "clear answer".

In the book Public Policy in an Uncertain World President Lyndon B. Johnson, when presented with a range of values for a forecast, is claimed to have said

Ranges are for cattle. Give me a number.

There’s a lot of low-hanging fruit for companies keen to embrace data science.

Absence of evidence is not evidence of absence

By

A statistical note published in the BMJ (formally the British Medical Journal) reminds us that

…trials that do not show a significant difference between the treatments being compared are often called "negative." This term wrongly implies that the study has shown that there is no difference, whereas usually all that has been shown is an absence of evidence of a difference. These are quite different statements.