Visualization Code for the Beginner Data Scientist

When I started as a data science student, I really started at zero. No coding experience. I’d long forgotten any of the higher level math that I knew. Towards the end of the first project at my data science bootcamp, I’d struggled with the code, struggled with modeling and struggled with the statistical underpinnings only to realize, with a few hours left before I presented my findings, that I hadn’t really visualized anything.

If you’re relatively new to data science, then you may not be aware of how crucial visualizations of your work can be in the real world. Depending on your role, a huge portion of your role may simply be the ability to communicate why your work is important. If the decision makers at your company don’t see how your work is valuable or actionable, then ultimately your work may be fairly pointless. The best way to communicate that value clearly, is to visualize it.

As I learned of course, it’s easier said than done, especially when you’re in the middle of learning so many new things. So to help out, here is some useful visualization code I learned through my bootcamp experience.

The below libraries will cover all the functions:

A correlation heatmap

A personal favorite. In this case we’re using a heatmap to easily identify which features of this model are highly correlated to the target (and how correlated they are to each, so we can deal with multicollinearity). Also the color scheme makes me feel like I’m walking away from an explosion.

KDE/Histogram

Great for checking if your linear features are normal distributions or not.

Confusion Matrix

This one is adapted straight from the scikit-learn documentation. Great for classification projects.

Related posts:

When you are born into a reality where choice is restricted and choicelessness is conditioned into the tapestry of its culture, you are faced with a choice: Do you stay and accept the conditions that…

The exhibition presented several works of art created with different techniques. The spiral and circular forms of the artworks have been inspired by the geometry of the traditional and classical arts…

In the previous blog I promised to talk a bit more about what I’ve learned here and how I see my future after this process. When I arrived here I knew I had a problem to control my use of drugs and…