Seeking to absorb the good from others doesn’t make our contribution any less. It just makes it all work better together.

Someone taught me about the relationship between the Tonalism art movement and the topic of unity this past Sunday. Tonalism was an attempt of “painting musical compositions”, making soft landscapes in the 1880’s. The method was to take a color (gray) and mix it in with a soft, carefully chosen color palette. Taking this one toner and blending it with all other colors brought unity and softness to the palette.

--

--

Peter Levine’s article, “The End of Cloud Computing” begins with an innovative question:

Subtract something that’s important today and fill it with something else, and you’ll start to think out of the box, as opposed to sequentially.

His article is inspiring. The future for data science, math, statistics, machine learning is just beginning. Not (specifically) in the AI-will-dominate-us sense, but in the nothing-won’t-have-ML-and-sensors sense.

Big trends:

  • Geospatial time series data
  • Sensor data (logs data)
  • On-device ML training and prediction
  • Even more connected devices
  • Less logic coding, more inference and prediction

This implies that data will design software, rather than coders designing software.

Yes, your umbrella, power drill, pillow, bed, shoes, [whatever], etc. will soon have sensors in it.

--

--

Reading a chart is reading data. We’re subjected to the angle of vision of the chart’s author. What data is included? What’s excluded? Reading a chart is just like reading a paragraph. A good chart is simple. But sometimes it’s beneficial to tell a robust story in the chart that a simple chart can’t tell.

  • Read the title: what am I looking at?
  • Look at the legend: what is being plotted or compared?
  • What is the line? Average? Who does it represent?
  • Look at the aces: what’s the magnitude of the change? Range?
  • Look at the trend. Up, down?

Takeaways/deductions:

  • Why is the trend this way? Do we like the trend? Do we want to change it?
  • What does this imply? So what? How many people are affected by this?

Questions to ask/critical thinking:

  • Year over year? Compare to last year.
  • Compare to benchmark?

--

--

You won’t know until months later what you wish you had logged.

There’s a trade off. The more you log, the slower you system is and potentially the more bugs you can introduce. The less you log the less likely you are to catch bugs.

Here’s a rule of thumb: log more than you think you’ll need. Then pare back.

No company has infinite resources.

--

--

Bryan Whiting

Bryan Whiting

The world is defined by writers | Silicon Valley Data Scientist | Google