Rating Engine Optimisation for E-commerce

CZ1115 Introduction to Data Science & Artificial Intelligence
Academic Year 2020/2021 Semester 2
Nanyang Technological University

---

Objectives

• To study how various marketing and customer service factors could influence customer satisfaction on e-commerce sites, as reflected by product ratings.
• To develop a model that is useful for retailers to improve their marketing strategies and deliver quality customer service, ultimately improving their product ratings.

In particular, the following predictor and response variables were studied:

Predictor Variables

1. Delivery Time
2. Deviation from Estimated Delivery Date
3. Length of Product Name
4. Length of Product Description
5. Number of Product Photos
6. Freight Value

Response Variable

1. Product Rating

The following supervised machine learning methods were used to develop our model:

Machine Learning

• Univariate and multivariate decision trees
• Random forest

Dataset

Brazilian E-Commerce Public Dataset by Olist

Contributors

• Tan Xin Kai
• Wong Yi Pun
• Wu Jun Hui