PREML: Photometric Redshift Estimation Using Machine Learning

📘 Photometric Redshift Estimation using Machine Learning [PREML]

This repository contains data for the Photometric Redshift Estimation Machine Learning PREML Dataset, which explores the use of machine learning models for estimating redshifts from photometric data and images. The project was carried out in two phases:

Comparative analysis of baseline ML models

A PGNN-based multimodal redshift estimation pipeline

📁 File Structure

PREML/
├── Photometric Data.zip
│ ├── HSC_PDR3/
│ │ └── galaxy_photometry.csv (816,743 Rows)
│ └── SDSS_DR18/
│ ├── galaxy_photometry.csv (2,790,253 Rows)
│ ├── quasar_photometry.csv (866,338 Rows)
│ └── star_photometry.csv (962,162 Rows)
│
├── HSC Photometry.zip
│ └── HSC Photometry.csv (395,585 Rows)
│
└── HSC Images.zip (395,585 files)
    ├── objid1.npy
    ├── objid2.npy └── ...

🧪 Datasets

Photometric Data

HSC_PDR3

galaxy_photometry.csv: Contains photometric magnitudes and spectroscopic redshifts for galaxies.

SDSS_DR18

galaxy_photometry.csv: Galaxy data

quasar_photometry.csv: Quasar data

star_photometry.csv: Star dataAll files include magnitudes and spectroscopic redshifts.

Multimodal Data

Uses both photometric and image data.

HSC Photometry.csv: Photometric data for objects with corresponding FITS images

HSC Images.zip: Contains .npy files named by object ID, each representing image data extracted from FITS files

📜 Citation

Ferrao, J., Dias, D., Naik, P., D'Cruz, G., & Naik, A. (2025). PREML: Photometric Redshift Estimation Using Machine Learning (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.15426394