top of page

Tableau実践問題集 #TableauChallenge を作りました。

Tableau Training 2-3: Trending Youtube Video Dashboard - Views and # of Videos by Categories


In the last post we preped the data we will use, so it's time to start creating our dashboard. Before starting it let's review what we have to identify and what to visualize.


1. What category we should focus on? As for metrics, we will visualize total views and number of categories within categories. 2. What channel we could use as our influencer. We will visualize its activeness that could be measure by how many videos they posted and how much views they earned within a certime period. 3. What video we should put our ads. We will visualize the number of views by videos. We may care if the number is glowing or stagnated.

Let's start with the first one because it could be helpful for us to have a big picture of the trending video circumstance. I would like to reccomend you to start with something comprehensive. In fact, under "Category" there are "channel_title" and there is "title" under the channel title, which is a title of a video. Considered that we start with the most comprehensive one and dive into details, where we could use filter/highlighter in Tableau, so far I believe that it's good to start with the top of the structure of the dimensions.

So let's start with observing the categories. By putting the Trending Date, Category and Views and Number of records, you will get this visualization below.


And now you can get some senses of what's going on. I set the individual axis range and it look like, for example, Films and Aniamtion is good and interestingly Autos&Vehicles had a peak. But since our KPI's for categories could be the total views within categories at the latest day because we may want people to see our ads now, we may want to know "what has the most views recently". In other words, we may want to sort the charts.

In order to only use the latest day, we have 2 options for this situation (charts). One is to use Table Calculation and the other is using LOD calculation. Let's see the both using the Number of Views.

Views at the latest day (with Table Calc) WINDOW_SUM(SUM([views]),LAST(),LAST())

Views at the latest day (with LOD) IF [Treding Date as Date]={ FIXED [Category]:MAX([Treding Date as Date])} THEN [views] END


The picture above is showing both calculations using the Entertainment category. The calc with LOD calculates the number is the Trending Date is the latest date. The calc with Table calc calculaters the number at the every data point, but it calculates the number at the latest day. This is the reason of the difference where one only has 1 data point and the other has continuous data points.

So by using it, let's sort the categories. The way to do this is different by which calc you use. I made videos about sorting so let's take a look them.




So now we've got a simple dashboard. However, this dashboard has some to be improved. ・Since Tableau doesn't have a synced scroll bar, it might be good to make the height fit the size of the dashboard. However, it makes a bit messy like the picture below.


・We may not need entire data so it'd be better to put a date filter, for example, we may only need last 10 days. ・Having 2 individual line charts is causing to use much space while the room for the dashboard is limited. It'd be better to try dual axis. ・And some format issue.

So we've got the basis of our dashboard. It's time to refine it. However, I don't go to how to create the finalized dashboard directly. I'd like to share my all tryal-and-errors through creating the dashboard, because I'd like to share why this is not good and why that is good I think. There's no right answer for a dashboard according to books, but I believe it's good to discuss what and why we think it's good. So you may feel it's long but let me state my opinion. You can also skip to what you want to know.

So let's deal with the first issue. So far we have line chart and the latest number to indicate the chart's scale. And it seems that the number makes it messy so let's delete them from the chart but put them next to the chart.

For the sake of better spacing, let's use "Dual Axis" to combine the views and number of videos.


(This is my first how-to video with my voice/English. I hope you would not feel uncomfortable much with it...)

At the video above, I explained how to make the dashboard less messy. However, that is not enough because it's still showing the number as just a number, which is not the way of data-viz. So the next video below I explained how to make a colored number list. This is a small tip but it's very useful. I learned it from this online course. I show you it as a reference.


(I mentioned that we will have the next video, but we'll not. To solve the alignment issue, I just forgot to check "Show Header" option for the bottom header for min1. You can hide the header just by right click on the space at the bottom.)

I left how to arrange the dashboard like mine to you. You can see how it is done with my dashboard here.

In the future post I may explain how to arrange the dashboard, but for today, since I put much information, let's end it here.

You may have some questions for the process we've walked through. Please send me a message through Linked-in or you can send me an e-mail through "Contact Me" page.

I hope you enjoyed this post. See you then.

Yoshi

bottom of page