Power BI Project: Analyzing Brazilian E-Commerce Public Dataset by Olist

Recently, I had a chance to work in a team on a very interesting Power BI project to analyze a Brazilian E-commerce Company — Olist. The dataset is published by Olist on Kaggle, and it has good quality, we barely did not have to clean the data before visualizing it. In my opinion, this is a good dataset for you if you wanted to practice and sharpen your Power BI skill.

There are a lot of things to explore in this dataset and I also learned various types of charts in Power BI after this project. In the video below, you can see the dashboard that I made to bring insight and from that, give some recommended improvements to Olist.

My Dashboard

From the first dashboard, you can have an overview of Olist’s performance in the duration of the dataset, from 10/2016 to 10/2018.

There are 3 main things I learn when created this dashboard:

• To create the card with various graphic designs, I need to download “Infographic Designer 1.9.5” in the option “Get more visuals” under the tab “Build visual” because that is not the default visual. There are also many other interesting visuals waiting for you to explore.
• To make the line forecast, I use the “Forecast” function that already has in the Analytics tab. Remember that not every visual has that forecast option, and if the dataset had any blank value, the forecast function can not work. So it is essential to exclude all the blank values from that visual by using Filters.

In this case, you can see how I modified the option in the below image. I had to “Ignore the last” 1 unit because the latest month is 10/2018 was not fully recorded, “Seasonality” is 12 — representing 12 months, forecast length is 14 months.

• To color the bar due to the number of orders, I used “Conditional Formatting”. I found it super useful when showing the relationship between 3 dimensions in the bar chart.

For example, in this chart, we can see the Average review score gradually decrease when the Delivery Duration increase by the bar’s color. And the best duration for delivery is below 20 days.

In the second dashboard, I focused on analyzing the review table.

Since all the review messages in the database are in Portuguese, I had to translate all of them into English by using Google spreadsheet online and the function =GoogleTranslate(E5, “auto”, “en”). This function will automatically recognize the language in cell E5 and translate it into English, nice and easy.

The Word Cloud visual is, again, not the default visual Power BI has, but you can easily get it in the “Get more visual” option. The bigger the word is, represent for the more times it repeats in the review messages that customers left. And we all can see that the Delivery service and Product quality are definitely what customers care about the most.

In this post, I will not mention what we — as a team — suggest for Olist to improve its business. I am sure you — based on your own visuals — can come up with many brilliant recommendations for improvement. Thank you and happy visualizing!