Cover by Q. Ethan McCallum

Safari, the world’s most comprehensive technology and business learning platform.

Find the exact information you need to solve a problem on the fly, or go deeper to master the technologies and skills you need to succeed

Start Free Trial

No credit card required

O'Reilly logo

Chapter 7. Will the Bad Data Please Stand Up?

Philipp K. Janert

Among hikers and climbers, they say that “there is no such thing as bad weather—only inappropriate clothing.” And as anybody who has spent some time outdoors can attest, it is often precisely trips undertaken under more challenging circumstances that lead to the most noteworthy memories. But one has to be willing to put oneself out there.

In a similar spirit, I don’t think there is really such a thing as “bad data”—only inappropriate approaches. To be sure, there are datasets that require more work (because of missing data, background noise, poor encoding, inconvenient file formats, and so on), but they don’t pose fundamental challenges. Given sufficient effort, these problems can be overcome, and there are useful techniques for handling such situations (like tricks for staying warm during a late-November hike).

But basically, that’s remaining within familiar territory. To discover new vistas, one has to be willing to follow an unmarked trail and see where it leads. Or equivalently, when working with data, one has to dare to have an opinion about where the data is leading and then check whether one was right about it. Note that this takes courage: it is far safer to merely describe what one sees, but doing so is missing a whole lot of action.

Let’s evaluate some trail reports. Later, we’ll regroup and see what lessons we have learned.

Example 1: Defect Reduction in Manufacturing

A manufacturing company had developed a rather ...

Find the exact information you need to solve a problem on the fly, or go deeper to master the technologies and skills you need to succeed

Start Free Trial

No credit card required