Dashboard Week
Dashboard week is an infamous week marked on every Data Schooler’s calendar where, for every day, you’re given the task of creating a new dashboard and writing a blog for that day. I missed my cohort’s Dashboard week, and have since been catching up whilst (attempting) to adhere to the strict one-day time block. Here’s my work for the second Dashboard week challenge, in which everyone had to create a Youtube Trending Dashboard. Click here to see my previous challenge.
Introduction
Day 2’s challenge involved using data that had information pertaining to Youtube’s trending videos of the last couple of years. The website had data across many different regions and so I decided to purely focus on Youtube in the US.
Data Preparation
There was little preparation to be done with this dataset. My intention was to see what kind of Youtube titles instigated more discussion, and so brought US trending data into Alteryx and performed a sentiment analysis on the title field.
Dashboard
My final dashboard covered whether sentiment affected things like a video’s comment per view ratio, as well as what categories of video were the most positive/negative.
I wanted to attempt to create a visually engaging thumbnail by plotting all of the videos on the axes below and colouring based on emotion, with a dark background making the colours ‘pop’. Because this kind of chart is visually appealing but does little to describe any insights, I then decided to bring in summary text for particular metrics as well as an actual description of the findings.
Because this first part was fairly simple, I decided to create a long dashboard and then focus on what video categories were the most positive and negative underneath the overall summary. Box plots were the best method for this, and allowed me to see if any averages were skewed due to outliers or small sample sizes. Again, having a written description alongside the chart helped with delivering insights.
You can find the final dashboard on my public Tableau profile.
