## Competition Description

Synthetic data is becoming a vital tool in modern data science, enabling research, innovation, and product development while preserving privacy. In time-series applications, generating synthetic sequences is especially valuable for domains like finance, healthcare, and IoT, where sensitive real data cannot be openly shared.

In this challenge, organized by **Genuity** in collaboration with **ETHOS Coding Club, IIT Guwahati**, your task is to build high-quality **synthetic time-series datasets** using our open-source library. The competition is part of ETHOS, IIT Guwahati’s annual coding fest, and is designed to give participants hands-on exposure to cutting-edge generative modeling techniques for tabular and temporal data.

### Problem details
- You are given a **multivariate time-series dataset** with equally spaced timestamps.  
- A **60% split** of this dataset is provided to you as `real_0.6.csv`.  
- The remaining **40%** is held back by the organizers and the evaluation done on whole dataset.
- Your objective is to generate a synthetic dataset of  3322 rows and same schema as the training set, using the functions available in the Genuity library.

A combined score will determine the leaderboard ranking. Public leaderboard scores are based on part of the hidden data, while final prizes are awarded based on private leaderboard scores.

### Why this matters
- **Research relevance:** Learn how synthetic data can safely unlock insights from time-series.  
- **Practical skills:** Work with a production-grade synthetic data library.  
- **Community impact:** Contribute to the growing ecosystem of tools for privacy-preserving data science.  

---

By the end of the competition, participants will not only showcase their ability to generate realistic synthetic time-series but also gain deeper insight into the scientific principles of generative modeling and evaluation.