BehaViewer

Graph of Picking Algo
Heroku Backend
Heroku Webserver
React Website for Business Purposes
Intro Page

Inspiration

In todays market, understanding client behavior is more important than ever when considering the massive impact of data driven decisions. We wanted to create a tool that allows organizations to gain deep insights into client habits, preferences, and potential future actions. This inspired us to develop BehaViewer, an intuitive and powerful application that visualizes user data and provides predictions on future behavior, helping organizations make proactive business decisions.

What it does

BehaViewer assesses user data, such as Internet Service Providers, Age, and device usage patterns, for actionable insights. By predicting trends in behavior, it helps organizations understand what their clients may do in the future. The front-end uses React and Tailwind CSS to provide a clean and user-friendly interface from which user activity is tracked then processed through the python based backend hosted on Heroku. For the sake of this initial phase, the website created mimics that of a cellular network provider like Telus, Rogers and Bell.

How we built it

The backend was completed through the creation of a machine learning algorithm written in Python using libraries like matplot lib, pandas, numpy and flask. The frontend was entirely completed through the use of React and Tailwind CSS.

Challenges we ran into

We’re incredibly proud of the robustness and flexibility of our algorithm, which successfully accommodates a wide variety of user data types. BehaViewer can handle diverse data points—from demographic information like age and gender to device preferences and service plan types—allowing organizations to leverage comprehensive insights into user behavior. Our HistGradientBoostingClassifier model is capable of processing and accurately predicting behavior across these varied data types, which enables a more holistic understanding of client actions and preferences. Balancing this algorithmic power with an intuitive front-end, we’ve built a tool that’s both versatile in data handling and accessible in user experience.

Accomplishments that we're proud of

One of the biggest challenges of this project was efficiently handling large amounts of JSON data and recognizing patterns within it to make meaningful predictions. Working with such complex and varied data required careful preprocessing and transformation to ensure our algorithm could interpret it effectively. On the technical side, integrating our front end and back end with the database posed additional hurdles. Ensuring smooth communication between the client interface and the server, while handling real-time data from our MongoDB database, required careful optimization to maintain performance and reliability across the application. Balancing these technical demands with a user-friendly design was both challenging and rewarding.

What we learned

Throughout the development of BehaViewer, we gained a deeper understanding of handling large datasets in JSON format to extract meaningful insights. While we faced challenges working with MongoDB, this experience highlighted the importance of efficient database management and data retrieval techniques, which we aim to improve in future projects. We also learned how to integrate machine learning models within a full-stack application, allowing real-time data processing and predictions. Working with React and data visualization libraries taught us valuable lessons in creating intuitive user interfaces for complex data. Overall, this project sharpened our technical skills and provided insights into building scalable, data-driven applications, while emphasizing areas for future growth in database management.

What's next for BehaViewer-FrontEnd-React

Next, we plan to enhance BehaViewer by incorporating a robust and optimized database integration, such as MongoDB, to streamline data handling and enable more scalable storage solutions. We also aim to improve our machine learning model to support a broader range of data types, allowing for even more accurate behavior predictions. On the front end, we’re hoping to explore advanced customization options, such as user-configurable dashboards and enhanced data filtering, so users can tailor insights to their specific needs. Additionally, we plan to implement user authentication and role-based access to make BehaViewer suitable for enterprise environments. These enhancements will make BehaViewer an even more powerful and flexible tool for organizations looking to understand and predict client behavior.