Kaggle Competitions Portfolio kaggle profile link : https://www.kaggle.com/vatsal2004gupta

Welcome! This is a living portfolio of Kaggle competitions tackled with reproducible notebooks, clear reasoning, and honest evaluation. Browse challenges, run the notebooks, and leave feedback—suggestions and PRs are encouraged.

Quick start Skim a competition write‑up to see the problem framing and approach.

Open a notebook and run it end‑to‑end to generate a submission.

Compare versions in the submissions history to see progress over time.

Explore competitions Prefer tabular problems? Check out Spaceship Titanic for feature engineering and stacking with tree‑boosting.

Interested in baselines? Look for “baseline” notebooks to see clean, minimal starting points before ensembling.

Curious about validation? Open CV notebooks where splits, metrics, and pitfalls are explained.

About me B.Tech CSE student (Amity University, India) focused on ML/DL, with hands‑on work in notebooks, pipelines, and reproducibility.

Enjoys open‑source, optimizer benchmarking, and turning experiments into readable, useful documentation.

How I work Start simple, then iterate: baseline → targeted features → CV diagnostics → ensembles only if they help.

Evidence‑driven decisions: show metrics, note trade‑offs, and keep seeds and environments pinned.

Share learnings: short ablations, error analyses, and notes that others can reuse.

What you can do here Run a notebook and produce your own submission; tweak a feature or model and compare the lift.

Open an issue with a question or improvement idea—responses are welcome.

Fork the repo, try a different CV scheme or loss, and open a PR with findings.

Discussions What feature mattered most? Which CV split was most reliable? Add thoughts in issues, or suggest a mini‑study.

Want a walkthrough call or code pairing on a specific competition? Open an issue titled “Walkthrough request”.

Roadmap More data‑centric error analysis and leakage checks.

Lightweight experiment tracking for quick comparisons.

Focused ablation studies to separate signal from noise in features and ensembles.

Ethics and data No redistribution of competition datasets; pointers provided to download via Kaggle.

Public results and methods are credited where used; please extend the same courtesy.

Stay in touch Open an issue for collaboration ideas or feedback.

If a notebook helped you, starring the repo or sharing improvements is appreciated.

Thanks for stopping by—happy modeling!