CXC Submission
Federato Submission
Federato Outline
- Objective: Enhance user engagement, reduce drop-off rates, and optimize user journeys on the Federato RiskOps platform by identifying key bottlenecks, user behaviors, and actionable insights using data-driven methodologies.
- Data Sources & Technologies: Data Format: JSON files converted to CSV, then ingested into DuckDB for efficient querying and analysis. Tools & Libraries: Python for data processing and analysis Pandas for data manipulation DuckDB for fast, in-memory SQL querying Matplotlib and Seaborn for data visualization NetworkX for user journey graphing and Markov Chain analysis Scikit-learn for clustering (K-Means) and propensity score modeling Plotly for interactive visualizations (where compatible)
- Methods Used: Exploratory Data Analysis (EDA): Analyzed user events, session durations, device types, and regional data to understand baseline engagement patterns. Clustering Analysis: Used K-Means clustering to segment users into groups (Inactive Users, Casual Users, Power Users, Outliers) based on interaction levels. Markov Chain Modeling: Built transition matrices to map user journeys and identify high drop-off points and inefficient navigation paths. Propensity Score Matching & Causal Inference: Assessed the impact of specific event categories on user engagement and identified key drivers of retention. Drop-Off & Bottleneck Analysis: Detected high-risk transitions and created targeted strategies to reduce user drop-offs. Relative Risk Analysis: Compared engagement levels across device types (Windows vs. Non-Windows) to uncover platform performance discrepancies.
Built With
- jupter-notebook
- python
Log in or sign up for Devpost to join the conversation.