- Please check a jupyter book or PDF file under notebook for data analysis.
- This data analysis project was introduced during the coding bootcamp.
- It was a half-guided module and the problem and the preconditions are presented.
- The insights, the visualizations and the suggestions are my original.
-
Brazilian E-Commerce Public Dataset by Olist
-
kaggle.com/olistbr/brazilian-ecommerce
-
Olist offers Logistic and Inventory Management Service to sellers
-
Information about ~100k orders made between 2016 and 2018
-
8 csv files (orders, cutomers, reviews, sellers, products...etc)
How should Olist improve it's profit margin, given that the revenue and the cost are calculated as the following condition:
Revenue
- Olist takes a 10% cut on the product price (excl. freight) of each order delivered.
- Olist charges 80 BRL by month per seller.
Cost
- Estimated cost occured by bad review per order
| review_score | cost (BRL) |
|---|---|
| 1 star | 100 |
| 2 stars | 50 |
| 3 stars | 40 |
| 4 stars | 0 |
| 5 stars | 0 |
- IT costss
Olist's IT costs are estimated to be proportional to the square-root of the total cumlated number of orders approved.
The IT department also told you that since the birth of the marketplace, cumulated IT costs have amounted to 500,000 BRL.
After the data analysis on seller data, I would give the following 2 suggestions to improve the profits by at least 10 %.
1. By removing the worst 15 sellers (0.5 % of the sellers) who make negative profits, Olist improves the profits by 10 %
2. By charging 10 % of the review cost directly to the sellers, Olist improves the profits by 13.9 %