Do Attributes such as Review Scores Affect the Sale of Products? — an EDA on Olist
What is Olist?
Olist is an online e-commerce platform founded in 2015 in Brazil. Olist connects merchants and their products to main marketplaces and they have developed a platform for shopkeepers of all sizes and segments to register their products to be sold at Olist.
EDA Goal
Ascertain if key attributes of the business processes drive high sales of products.
A Sub-Question
What factors predict sales outcome; what is the importance of review scores as compared with other factors?
Insight and Considerations:
- Order status is categorical data — it cannot be used here
- Only review score and freight value have positive strengths
Key Findings
- Payment sequential, payment installments, and freight value correlate to sales outcome but do not predict it — higher priced items tend to have higher shipping values because they involve either premium shipping or are heavy items
- Review scores do predict sales outcome
Are the sale of products affected after positive/negative reviews? Do negative reviews result in loss of customers?
Insight: As revenue generally decreased from 2017 to 2018, negative reviews have an effect on revenue.
Insight: Revenue and the ratio of good reviews increased, indicating positive review scores affect revenue
Conclusion
Negative review scores would result in more negative review scores in the future (same trend with positive review scores). Additionally, products with more negative review scores see a decrease in revenue and products with more positive review scores see an increase in revenue.
Github Repository: https://github.com/MiloSta1n/Exploratory-Data-Projects/blob/main/Sean_Tay_EDA_Project.ipynb