# GDGC AI/ML Inductions Batch 2025-26' — Data

来源：https://www.kaggle.com/competitions/gdgc-ai-ml-inductions-batch-2025-26/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

# Dataset Overview

This dataset is designed for a supervised learning task where the goal is to predict the probability of a relationship based on encoded features. It includes training and test sets, a feature metadata file, and a sample submission template.

---

## Files Provided

| File Name               | Description                                                                 |
|------------------------|-----------------------------------------------------------------------------|
| `train.csv`            | Training data with 5,000 samples, 33 encoded features (`F1`–`F33`), `ID`, and `target` |
| `test.csv`             | Test data with 2,000 samples, 33 encoded features (`F1`–`F33`) and `ID` only |
| `feature_lookup.csv`   | Maps encoded features (`F1`, `F2`, ...) to actual names and type (numeric/categorical) |
| `sample_submission.csv`| Template showing the required format for submission: `ID` and `relationship_probability` columns |

---

## Column Details

- **Encoded Features (`F1`–`F33`)**: Anonymized input features used for prediction.
- **ID**: Unique identifier for each sample.
- **target** (`train.csv` only): The target variable representing the probability of a relationship.
- **relationship_probability** (`sample_submission.csv`): The predicted value to be submitted for each test sample.

---

##  Feature Metadata (`feature_lookup.csv`)

This file provides additional context for each encoded feature:

- **Feature Name**: Actual name corresponding to each encoded feature (`F1`, `F2`, etc.).
- **Feature Type**: Indicates whether the feature is `numeric` or `categorical`.