# Global AI Hackathon'25 by Elucidata — Data

来源：https://www.kaggle.com/competitions/el-hackathon-2025/data

采集时间：2026-10-19T09:10:39Z

All the data need to train model and make predictions is bundled in a single .h5 format file (elucidata_ai_challenge_data.h5). The data has been divided into Train and Test set and the file consists of the following objects:

Train:

- Spots: The 2D coordinates of the spots in the Train set slides (n = 6 slides); this is a dictionary of tables with the Train slide identifiers S_1 to S_6 as keys. Each table provides the (x,y) coordinates and the assigned cell type abundances (35 cell types) as one row per spot. 
- Images: The aligned histology images for the Train set slides, as 3D numpy arrays (n = 6 slides); this is a dictionary of tables with the slide identifiers S_1 to S_6 as keys.

Test:

- Spots: The 2D coordinates of the spots in the Test set slide (n = 1 slide); it is formatted as a dictionary with key being the identifier of the test slide (S_7) and the spot coordinates provided as a table, with one spot per row. The cell type abundances of the Test slide spots are to be predicted.
- Images: The aligned histology image for the Test slide S_7, as 3D numpy array (n = 1 slide). for consistency with the Train set, it is formatted as a dictionary with the key being the Test slide identifier (S_7) and value being the numpy array corresponding to the Test slide HE color image .

Please refer to the shared Starter Notebook for details on accessing the contents of the .h5 data file and visualizing the Train and Test data.

## Files

*   **elucidata_ai_challenge_data.h5** - the training and test sets
*   **sample_submission.csv** - a sample submission file in the correct format
*   **starter-notebook-kaggle.ipynb** - a python notebook to load the data in the .h5 file, explore the training data, and prepare a submission file for evaluation (https://www.kaggle.com/code/elucidata2025/kaggle-starter-notebook)