From the course: 15 Mistakes to Avoid in Data Science
Unlock the full course today
Join today to access over 22,600 courses taught by industry experts or purchase this course individually.
Having a data set that is too small
From the course: 15 Mistakes to Avoid in Data Science
Having a data set that is too small
- My name is Louis Tremblay. I'm a Senior System Engineer and I work at FLIR Systems. A common mistake that I often see is kind of using data that's not ready. And not necessarily just data that's not ready, but not enough data. And that can be a real challenge for a data scientist when you're presented with too small of a dataset or too small of a benchmark, you're going to be spinning your wheels trying to make things work. When in reality, the number one thing that you could do, the number one thing that you can improve your day-to-day efficiency is actually just spending the time to go get more data. So I actually have an example in my job where our team was working on a self-driving car dataset. And we started off with a couple hundred video clips, but ultimately, it wasn't enough. We were spending so much time trying to improve it. And the way I solved it, I just grabbed a camera and went driving for several hours.…
Contents
-
-
-
Communicating with overly technical language1m
-
Skipping the fundamentals1m 5s
-
Moving too quickly56s
-
Having a data set that is too small1m
-
Failing to adopt new tools1m 16s
-
Not considering the level of variation1m 20s
-
Lack of documentation1m 30s
-
Relying solely on formal education1m 22s
-
Taking too long to share results1m 10s
-
Including your bias1m 1s
-
Overpromising solutions to stakeholders1m 4s
-
Building tools from scratch1m
-
Assuming the knowledge level of stakeholders41s
-
Not telling a story with the data1m 53s
-
Not confirming with stakeholders1m 57s
-
-