Sleep Health A/B Testing Project

A data science project analyzing the impact of sleep duration on health outcomes using real-world lifestyle data. We conduct exploratory data analysis (EDA) and apply A/B testing to investigate correlations and statistically significant differences across groups.

---

Project Summary

• Cleaned and analyzed a dataset of 374 individuals
• Explored trends using histograms, pie charts, and correlation heatmaps
• Applied Welch's t-test to compare health outcomes between:
• Treatment group: Sleep ≥ median hours
• Control group: Sleep < median hours

---

Sample EDA Visuals

Sleep Duration Distribution

Stress Level Distribution

Sleep Disorder Category Breakdown

Correlation Matrix

---

A/B Testing

Participants were split based on median sleep duration:

• Treatment Group: Sleep ≥ median
• Control Group: Sleep < median

We tested whether sleep duration affects:

• Stress Level
• Heart Rate
• Quality of Sleep
• Physical Activity Level

Results were statistically tested using Welch's t-test.

Why Welch's T-Test?

We chose Welch's t-test over the standard Student's t-test because:

• Unequal sample sizes between treatment and control groups
• Different variances across health outcome variables
• Welch’s t-test does not assume equal variance or group size
• More robust for real-world, observational data like ours

---

A/B Test Results

We split the dataset into two groups based on median sleep duration and ran Welch’s t-test on each key metric:

Metric	T-statistic	P-value	Significant
Stress Level	-16.299	0.0000	Yes
Heart Rate	-7.962	0.0000	Yes
Quality of Sleep	23.309	0.0000	Yes
Physical Activity Level	7.068	0.0000	Yes

---

A/B Test Impact Ranking

I ranked each metric based on the absolute value of the t-statistic — the higher the value, the stronger the observed effect from increased sleep duration.

Rank	Metric	T-statistic	Interpretation
1	Quality of Sleep	23.309	Most impacted — people who sleep more report dramatically higher sleep quality.
2	Stress Level	-16.299	Strong inverse relationship — more sleep means lower stress.
3	Heart Rate	-7.962	Moderate effect — better sleep linked to lower resting heart rate.
4	Physical Activity Level	7.068	Still significant — slightly higher physical activity with more sleep.

---

Insights

• Sleep Quality shows the strongest positive effect from longer sleep duration.
• Stress Levels notably decrease with more sleep — highlighting mental health benefits.
• Heart Rate and Activity Level also benefit, but with comparatively smaller effects.

---

Run It Yourself

1. Clone the repo:

   git clone https://github.com/yourname/sleep-ab-test.git
   cd sleep-ab-test
   

2. Add your kaggle.json file to the root.

3. Install dependencies (recommended: uv):

   uv venv
   uv pip install -r pyproject.toml
   

4. Run data download script:

   python src/download_data.py
   

Dataset

Sleep Health and Lifestyle Dataset – Kaggle