To the friends that follow the blog, I would like to present some types of charts that should've been more explored on BI projects in general. Very often, they are left for more specific scenarios, or are not even considered because they look a little bit more complex for data demonstration.
It is true that many clients are more used to the traditional bar/column/line charts, or to the famous pie chart, and with good reason, after all, they are easily visualized and answer most business questions addressed to them. Even so, we can often try to enrich the front-end of a BI project with some visualization suggests that can be very useful during data analysis. In order to keep the text short, in this post we will consider 2 charts types that I consider under explored, and than, depending on the receptivity of it, we will show other possibilities on the following weeks.
Note: The charts are available in many BI tools and it is possible that the names differ among them.
Bubble Chart
This chart type is well known, although not always used. It is suitable for displaying three dimensions of data in one single crossing, being this a differential in relation to the bar/column/line charts.
We are going to demonstrate it with a very simple example easy to understand. We will use as example a cars re-seller.
We suggest a visualization like this:
Notice that in the same chart, the client can compare the sales percentage of each car (size of the indicator), the amount sold (x-axis) and the profitability achieved (y-axis)
It is very likely that the user already has a pie chart to show the proportion of sales and one or two column/bar/line charts to show the other two indicators. We were able to reduce all the view to one chart, keeping the simplicity in the data analysis.
And if the client misses the other traditional chart formats in this analysis, we can suggest a drill and detail the information in a timeline or with other relevant comparisons, as in this case, that the client can click on a bubble in the chart and get a detailed information in other three "traditional" charts:
Yes, in the end, we are back to the more traditional charts, but we created a initial point for the analysis, that can guide the user during information navigation and analysis.
Let's talk about other type of chart.
Tree Map
This chart can be very interesting to show a structured information, being able to present more than one level of a hierarchy as well as different dimensions of data.
In our simple example, following the logic presented above, we also have the performance information of a cars re-seller, but now not specific to one category, but for as many as needed (be careful with "as many as", the chart must be simple to be useful).
The chart presents 4 car categories (popular, medium, SUV and sports), being the size of each section the amount of cars sold in each category. It could have been sales region, brands, sellers or any other aggregated information. Within each category, we have new frames for each car model, that also have their amount sold represented by the size of it. The color intensity can yet represent another indicator, in this case, the profitability by car model (the lighter ones mean higher profitability).
In short, the user can analyse the sales proportion in two different levels (category and model) and also the profitability for each model. The same information, which maybe needed a few traditional charts to be demonstrated, is summarized in just one, keeping the simplicity to the analysis.
So, why couldn't we use this chart, like in the first example, as the initial point for a orientated navigation?
In short, the user can analyse the sales proportion in two different levels (category and model) and also the profitability for each model. The same information, which maybe needed a few traditional charts to be demonstrated, is summarized in just one, keeping the simplicity to the analysis.
So, why couldn't we use this chart, like in the first example, as the initial point for a orientated navigation?
It would be interesting if the readers could comment other possibilities for these or other "non-traditional" charts, as well as other interesting ways to improve the front-end in BI projects.
_______
Igor Alexandre Jakuboski is a Business Intelligence
professional for more than 7 years, he has worked in two of the main players on
the market for BI solutions and he has been leader in projects in Brazil and
abroad. Presently he is Principal Solution Architect at SAP.
Nenhum comentário:
Postar um comentário