Project title: Building and testing of a recommender system using a user based collaborative filtering technique.

Author: Tola Ogunniyi

---

Date completed: April 11, 2020.

Introduction:

One of the biggest challenge every E-commerce business faces is knowing what products to recommend to customers. This is very crucial because recommending the right product(s) helps to increase customer retention, trust and profit generation. There are different types of recommender systems and the one explored in this project is a user-based collaborative filtering technique.This recommends items that similar users have also liked i.e. predicting unknown ratings by using similarities between users.

Dataset

I used an Amazon fine-food-reviews dataset that I found on Kaggle.

---

Jupyter notebook

The project was completed using a jupyter notebook that consists of 2 parts viz:

Part 1: (i) Preprocessing and cleaning (ii) Exploratory data analysis

Part 2: Modeliing

Visualizations in the dataset

There are three figures provided. These are visualizations created in the exploratory data analysis (section ii) of Part 1. All three visualizations were created using Plotly express. The charts have also been included as an image file with links attached.A summary of the each chart is provided below:

Figure 1-Distribution of Scores for Amazon fine food reviews

Figure 2-Distribution of number of times users have assigned a score(s) to different products

Figure 3-Distribution of number of customers that have assigned a score(s) to different products

References

• https://towardsdatascience.com/building-and-testing-recommender-systems-with-surprise-step-by-step-d4ba702ef80b

Future work based on the dataset

• Next step would be to explore the dataset with an item-based collaborative filtering techniques and also deep learning to help in comparing the different models to see which performs the best.

---

Thank you very much for taking the time to look at this project. Please feel free to contact me via email([email protected]) or linkedIn if you have any questions,comments or feedback