Code Release Guide

Step 1: Dataset Integration

The original dataset is not sorted chronologically but organized in time chunks. Integration is required for easier usage.
Run ./Pre-processing/make_llamps_dataset.ipynb

Step 2: LLM Ranking and Analysis

Use LLM to process the data, score user texts, and perform analysis.
Run ./llamps/Pre-processing/LLMprocess.py separately for each of the following four paths. Running them concurrently can improve efficiency:

• ./llamps/Pre-processing/llamps-dataset/negative_examples_anonymous
• ./llamps/Pre-processing/llamps-dataset/negative_examples_test
• ./llamps/Pre-processing/llamps-dataset/positive_examples_anonymous
• ./llamps/Pre-processing/llamps-dataset/positive_examples_test

Note: Modify the url and api_key in GPTRank.py and GPTAnalyze.py. These are the interfaces for the LLM.

Step 3: Dataset Preparation

Run ./llamps/Pre-processing/process_sentence_embedding.py
Run ./llamps/Pre-processing/train_fast_dataset.py

Step 4: Training

Execute the following command in the terminal:
bash runme_combine16.sh

Step 5: ERDE Calculation and Visualization

ERDE Score Calculation

• File: llamps/erisk_infer_analyze.ipynb
• Calculate the ERDE (Early Risk Detection Error) metric for model evaluation

Sleep Pattern Visualization

• File: llamps/schedule_visulize.ipynb
• Visualize and analyze sleep characteristics and patterns from the data

Dimensionality Reduction Visualization

• File: llamps/t-sne.ipynb
• Apply t-SNE for dimensionality reduction and visualize high-dimensional data patterns