RecSys (Search)

Data is my inspiration every day its increasing at exponential rate which means we will have enough data to play with and learn from it. I choose recommendation systems which help customer engagement and smoothens digital experience.

This system will allow digital marketers to offers customers product and service recommendations in real-time. It engages with customers by personalizing choices and improving user experience.

There are two scenarios to consider :

1. Building a model based on collabrative filtering system.(Using previous purchase history)

2. Building a RecSys for new customers without any purchase history.

This system is built using python and few libraries such as pandas, sklearn etc.

During the martix formation the data was very large and took longer time for execution. It was difficult deciding the number of cluster using elbow method.

I was able to successfully complete the project with good working system and accuracy.

I was able to learn the decomposing and vectorization process deeply.

Next is scraping out the insurance search data using python script and generating dataset. Which can be used to modify the current recommendation system which allows to create customized policies and offers based on each individual searches.

Built With

• github
• pandas
• python
• scikit-learn