Competitions

ADIA Lab Structural Break Challenge — Top 7 / 1,842

ROC AUC Score: 88.34%

The ADIA Lab Structural Break Challenge, also known as “Can You Predict the Break?”, focuses on detecting structural changes in univariate time series data. A structural break occurs when the process generating a time series changes at a specific point, altering its statistical behavior.

The goal of this challenge was to design an algorithm capable of determining whether a structural break had occurred — by analyzing patterns in the data before and after a specified time point. Participants were provided with labeled training data to develop and calibrate predictive models.

Structural break detection plays a vital role in various domains such as:

• Climatology — identifying shifts in weather or climate trends.
• Industrial Systems — detecting anomalies that may indicate machine failures.
• Finance — recognizing regime changes in asset prices and market behavior.
• Healthcare — detecting sudden physiological signal changes for early alerts.

I developed a hybrid feature-engineering and machine learning pipeline that leveraged signal processing, statistical features, and CatBoost-based modeling. Through this approach, I achieved a top 7 ranking worldwide out of 1,842 participants with an ROC AUC score of 88.34%.