Sleep diary analysis

This repository contains a longitudinal sleep diary project that combines self-tracked sleep records with nightly bedroom sensor data on CO2, temperature, and humidity.

The aim is to study how behavioral, temporal, and environmental factors are associated with night-to-night variation in sleep duration and insomnia-related outcomes. The repository is organized as a reproducible analysis workflow, from raw data cleaning to descriptive analysis and regression modeling.

Study design

The project is based on within-person longitudinal data. The analytical workflow combines:

• sleep diary entries
• nightly environmental sensor measurements
• reproducible data cleaning and variable construction
• descriptive visualization
• regression-based modeling

The empirical focus is on variation across nights rather than differences between individuals.

Repository structure

• data/raw/: raw input files
• scripts/: data cleaning, joining, descriptive, and modeling scripts
• figures/: generated figures
• outputs/: model summaries and other analysis outputs
• archive/: older exploratory material
• projects/: additional project-specific work

Analysis workflow

Run scripts in numeric order.

1. scripts/01_load_main_data.R
   Loads and cleans the sleep diary data.

2. scripts/02_load_co2_temp_data.R
   Loads and aggregates nightly sensor data.

3. scripts/03_join_relevant_data.R
   Joins sleep diary data with nightly sensor summaries.

4. scripts/04_descriptives_and_plots.R
   Produces descriptive summaries and figures.

5. scripts/05_models.R
   Fits baseline regression models for sleep duration and insomnia.

6. scripts/06_fixed_effects_models.R
   Extends the modeling strategy with time fixed effects and sensor-based specifications.

7. scripts/99_smoke_test.R
   Runs a lightweight validation check after code changes.

Methods currently implemented

The repository currently includes:

• cleaning and harmonizing raw sleep diary data
• parsing and aggregating nightly sensor observations
• date-based joining of diary and sensor data
• descriptive analysis of sleep patterns
• baseline regression models
• time fixed-effects models for within-series analysis

Reproducibility

The repository is designed as a script-based workflow. Raw data are treated as immutable, and analysis-ready objects are created through the scripts in scripts/.

Because the project uses personal tracking data, raw input files may not always be suitable for public redistribution. The code structure is therefore intended to keep the analytical workflow transparent and reproducible even when data access is restricted.

Current status

The repository currently supports:

• data preparation
• descriptive analysis
• baseline modeling
• fixed-effects extensions for within-person inference

Ongoing work focuses on improving model specification, validation, and presentation of results.

Author

Johannes Piipponen

Quantitative research workflow for sleep diary and environmental data analysis.