Tuesday, November 05, 2013

What Data Scientists Can Learn From a Fourth Grade Teacher

Fourth Grade Lesson on Data Science
Last week, I attended my fourth grade son's open house at school and sat through his math class. The teacher walked through the steps to solve a word-math problem. You know the kind - "Sally has $3.85, spends $1.20 on materials for lemonade then sells four cups at $0.65. How much money does she have?"

This was a great experience for me to observe. I am sometimes critical of US schools and generally feel that they need to do a better job preparing students, especially in math and science, in order for them to better compete in a global marketplace for skills and talent.

The teacher was on the right track by showing kids how to think methodically and solve problems. She completed several word-math problems, calling kids out to help her go through each step one at a time. I am certain that more kids "got it" because the teacher helped her students think, repeated by example, and required kids to show their work along with the answer.

It was very interesting to see the teacher discuss "data" with kids. Did I learn about data in fourth grade? I doubt it. I wonder if it's lessons like these that are early preparations for some of them to become the next generation of data scientists. Bravo!

Lessons for Today's Data Scientists

This fourth grade lesson is applicable to today's scientist. My interpretation is below with italics showing the fourth grade approach and bold showing the data scientist methodology.

  1. Slowly read the story - Data scientists have to attentively listen to their customers, colleagues, and others to help get a sense of what types of analytics are valuable, what decisions they will drive, and how discovery can lead to action.
  2. Understand the question - Sometimes, an executive will give the Data Scientist a list of questions, but more often than not, the Data Scientist has to be adept to ask good questions. (Note - see my older post on Asking Smart Big Data Questions.)
  3. Circle the important data - Visualize the most relevant data and apply algorithms to help discover data relationships and correlations. Use the visualizations to develop stories and tools that drive and backup insights.
  4. Check your response - Amazing how many times analysts deliver insights out of context. The data scientist needs to consider multiple approaches to insure the results are conclusive, or check with subject matter experts to insure results are reasonable.
Most important - possibly the subject of another post - Data Scientists need to "show their work". Data driven insights need to be backed up with views of the data, the algorithms, and the assumptions.





1 comment: