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.
Not considering the level of variation
From the course: 15 Mistakes to Avoid in Data Science
Not considering the level of variation
- A common mistake is also to not give enough weight to the amount of variation in your data set when you're trying to model or predict an outcome. For example, I recently worked on a project where I was using a pretty limited data set to predict how schools may perform on a summit of assessment, but the underlying data in that data set had a lot of variation across schools. Students could perform really well as a group or really poorly as a group, but it was really hard to use that data to make any predictions, because the amount of error of those predictions was very high. And that information is really difficult to convey to stakeholders in a way that's meaningful, because they would like to use the data to make decisions. And keeping in mind variation when making decisions is really difficult to do, and so it's often dismissed as unimportant because the decision has to be made, but the variation within the dataset…
Contents
-
-
-
(Locked)
Communicating with overly technical language1m
-
(Locked)
Skipping the fundamentals1m 5s
-
(Locked)
Moving too quickly56s
-
(Locked)
Having a data set that is too small1m
-
(Locked)
Failing to adopt new tools1m 16s
-
(Locked)
Not considering the level of variation1m 20s
-
(Locked)
Lack of documentation1m 30s
-
(Locked)
Relying solely on formal education1m 22s
-
(Locked)
Taking too long to share results1m 10s
-
(Locked)
Including your bias1m 1s
-
(Locked)
Overpromising solutions to stakeholders1m 4s
-
(Locked)
Building tools from scratch1m
-
(Locked)
Assuming the knowledge level of stakeholders41s
-
(Locked)
Not telling a story with the data1m 53s
-
(Locked)
Not confirming with stakeholders1m 57s
-
(Locked)
-