November, 2022

Olist - E-commerce optimization

Have a look at one of my early data analysis kaggle challenges. The goal is to increase the profit margin of an e-commerce platform based on customer and supplier data.

This project was based on the publicly available dataset of the brazilian e-commerce store Olist. The goal of the project is to explore the dataset and improve revenue by providing data-driven suggestions to the company. As part of the project, I preprocessed the data and transformed it to processible dataframes. Consequently, I conducted exploratory analyses, regression analyses on the most influential factors for negative reviews and worked out actionable suggestions for the company with calculatable profit margin improvements. The main insights generated in this process are as follows:

  • `Wait_time` is the most significant factor behind low review scores
  • `Wait_time` is made up of seller's `delay_to_carrier` + `carrier_delivery_time`.
  • The latter being outside of Olist's direct control, improving it is not a quick-win recommendation
  • On the contrary, a better selection of `sellers` can positively impact the `delay_to_carrier` and reduce the number of bad `review_scores` on Olist.
  • Dropping the worst performing 746 sellers from the dataset results in the highest profit margin for Olist, with a profit of ~1.2 million BRL
  • However, the profit increase already starts stagnating when around 200 sellers are dropped.


    • Considering the generated insights, the main recommendation for Olist is to drop the worst performing 200 sellers from the e-commerce platform.

    Thank you for the interest in this project. Feel free to inspect the github repository here.

      Tech stack

    • Pandas
    • Numpy
    • Matplotlib
    • Seaborn
    • Statsmodels