CxC 2025: Federato Challenge - Feature Farmers

Sequence Action Predictor

What it does

Our project includes an analysis of the Federato platform's user data, and a predictive model that is able to inference, with 80% accuracy, the next step a user should take given a sequence of 8 past events and user features (user role, city, country, device, language).

How we built it

We built our project after conducting data processing and using predictive modelling. We leveraged technologies such as PyTorch, scikit-learn and NetworkX, as well as Pandas, NumPy for data manipulation and processing, and Plotly, Matplotlib and Seaborn to create interactive and insightful visualizations.

Challenges we ran into

One challenge that we ran into was dealing with a large volume of data and identifying which columns and records are most important to our analysis. Our work became clearer after we were able to identify the key features in our data that we wanted to hone in on.

Accomplishments that we're proud of

We are proud of the overall outcome of our project, particularly the development of a predictive model that accurately forecasts the next step a user should take based on a sequence of past events and user features.

What we learned

This was our first time creating dashboards using Streamlit, familiarized ourselves with how to create data visualizations were both informative and user-friendly since clarity of data presentation is imperative in data science projects.