top of page

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

Tableau Training 2-4: Trending Youtube Video Dashboard - Influencer Analysis


In the last post we finished the category overview worksheet where we can easily see what category has the most views and the most number of the videos, which could be an indicator of how hot the category is.

Today, we will deal with what Youtube channel is most active and obtains many views, because we are looking for someone who can be our influencer.

Topic:

1. Gantt Bar with days

2. Dashboard Action

3. Tableau's Operation Order

So let's define how we measure the performance of Youtube channels.

Since we want our influencers to have trending videos frequently in order to show our Ads more, we have to measure how active they are. We also want to know how many views the Youtube channels earn because the total views directly affect how many times our Ads could be seen.

So we will set 2 KPI's here.

Activeness: How many trending videos they have within a certain days

Median Views: The Median Views of their videos within a certain days.

Let's see the 2 KPI's in Tableau.

Activeness

{ FIXED [Channel Title]:SUM([Number of Records])}

Median Views

{ FIXED [Channel Title]:MEDIAN([views])}

And we arrange the workbook in this way.


We have Date in the column and we have Channel Title, Category, Activeness and Median Views.

The channel title is sorted by Activeness.

Since we want to know when they have a trending video, we use "Gantt Bar" chart showing a Gantt Bar if there's a record for a day. I colored the bars with SUM(views) and I sized the bar with SUM(1). It's a personal preference and a small tip, but if you add the sizing SUM(1), it'd be better to show the sequence. Here we have a comparison of with/without SUM(1).

As for the color, I set 2,000,000 as the maximum because without doing it I couldn't see the difference with the most videos. It happens since some Youtube videos earn huge views but they're like outliers.

So I checked a boxplot of views.


According to it, the Upper Whisker is about 2,000,000 views then I used the number.

Then, by simply putting the worksheet in the dashboard we have, it will be like this.


It's obvious that it's messy with these points below:

1. We don't need all days. we need a date filter.

2. We want to see influencers in a certain category. We have to set dashboard action.

To deal with the 1st issue, I created a set with Trending Date as Date.


"How Many Days" is a integer parameter to set how many days we will show. I used the set as a filter and applied it to all the worksheets.

As for the second issue, let's use the dashboard action. The outline is like this: if we click the category at the left side, the line chart and bar chart next to it will be highlighted and the influencers will be filtered with the category we select. The setting will be like this.

Now you will see how they work.


You will see that clicking the category highlights the bar chart and line chart, and the influencers are filtered to show the related category. However, you will see one problem here. We sorted the influencers by Activeness, but it's obvious that, for example, Music Relaxing is less active than MigosVEVO, but it stands at the higher position.

The reason why it happens is because of Tableau's calculation order. According to Tableau's Order of Operation, it seems that FIXED calculation that we used for Activeness is done before the Date filter while we want to have Activeness within the date range.

To fix the problem, we set the Date filter as a context filter. Here is a before/after of context filter.

Now you see that the Activeness is calculated within the proper date range.

Then progress so far is like this.


So far, based on the data, we could say that the Music category has the highest views but doesn't have so many videos, which implies that each video has huge views, then we could say if our product is related to Music or our customer is interested in Music based on our customer demographic data, we can go into the category.

As for the influencer identification, we could say that based on the activeness of the channel and the median views, Speed Records and Marshmello could be good.

In the next post, we will identify what videos of the category or of the influencer are the best.

I hope you enjoyed this article. See you then.

Yoshi

bottom of page