Proposing the Roommate Sleep Preference Questionnaire (ROOMPREF) with a free online roommate matching tool

This paper introduces the Roommate Sleep Preference Questionnaire (ROOMPREF) and a corresponding free, Python-based web tool that utilizes adaptive K-Means clustering and a swapping algorithm to match roommates based on sleep preferences, aiming to reduce conflicts and improve sleep quality and performance in shared living environments.

The Gist

Imagine you are trying to organize a massive sleepover for 50 people. You have a few empty rooms, and you need to decide who sleeps with whom.

If you just throw names into a hat, chaos is almost guaranteed. You might end up pairing a night owl who wants to stay up until 3 AM with a early bird who needs to be asleep by 9 PM. Or, you might put a heavy snorer right next to someone who needs absolute silence to fall asleep. The result? Everyone is tired, grumpy, and performing poorly the next day.

This is exactly the problem the authors of this paper are trying to solve. They have created a simple solution called ROOMPREF (Roommate Sleep Preference Questionnaire) and a free online tool to fix it.

Here is a breakdown of how it works, using some everyday analogies:

1. The Problem: The "Sleep Mismatch"

Think of sleep like a custom-tailored suit. What fits one person perfectly might be a disaster for another.

• Chronotypes: Some people are like sunflowers (they wake up with the sun and go to sleep early), while others are like moths (they are most active at night). Putting a sunflower and a moth in the same room is like trying to force a cat and a dog to share a bed; they just have different rhythms.
• Environment: Some people need a tropical jungle (warm, dark, and quiet), while others are fine with a breezy, slightly noisy porch.

When these differences clash in a shared room (like a college dorm, a military barracks, or an Olympic village), it ruins everyone's sleep.

2. The Solution: The "Sleep Matchmaker"

The authors built a digital matchmaker to help coordinators (like dorm managers or team coaches) pair people up correctly.

Step 1: The Questionnaire (The "Sleep Resume")
Instead of guessing, everyone fills out a very short, 8-question survey (it takes less than 2 minutes). It asks simple things like:

• "Are you a morning person or a night owl?"
• "Do you snore?"
• "Do you need total silence, or can you sleep with some noise?"
• "Do you like your room hot or cold?"

Step 2: The Algorithm (The "Smart Sorting Hat")
Once the coordinator uploads the answers into the free online tool, the computer does the heavy lifting. It acts like a super-smart sorting machine:

• The Filter: It first separates people by gender (so men and women aren't mixed).
• The Grouping: It then groups people who are similar. It tries to keep all the "Morning People" together and all the "Night Owls" together.
• The Conflict Check: It has a special rule: "No Snorers with Silence-Seekers." If the computer sees a snorer and a person who needs silence, it tries to put them in different rooms. If it can't avoid it, it raises a red flag so the human manager knows to be careful.

Step 3: The Swap (The "Tug-of-War Fix")
Sometimes, the computer makes a mistake or runs out of perfect matches. The tool has a clever "swap" feature. If a snorer is accidentally paired with a light sleeper, the tool looks for another room where the roommate is a "heavy sleeper" (someone who doesn't mind noise) and swaps them. It's like a game of musical chairs where the computer rearranges the seats until everyone is comfortable.

3. Why This Matters

The authors tested this with a group of Paralympic athletes before the 2024 Paris Games. Before, they matched roommates based on who the athletes wanted to room with, which often ignored sleep habits.

With the new tool, they could create a sleep-friendly lineup.

• The Goal: To stop the "sleep wars" before they start.
• The Result: Better sleep leads to better performance. Whether you are an athlete trying to win a gold medal, a student trying to pass a math exam, or a soldier trying to stay alert, sleep is the foundation.

The Bottom Line

This paper isn't just about math or code; it's about human comfort. It's a free, easy-to-use toolkit that says: "Let's stop guessing who sleeps well together. Let's use science to make sure everyone gets their best rest."

It turns the chaotic process of assigning roommates into a smooth, logical process, ensuring that when you close your eyes, you aren't worrying about your roommate's snoring or their 3 AM phone calls.

Technical Summary

1. Problem Statement

Shared sleeping quarters are ubiquitous in high-stakes environments such as elite sports (Olympic/Paralympic villages), military barracks, and educational dormitories. However, mismatched sleep preferences between roommates frequently lead to sleep disturbances, which negatively impact human health, cognitive performance, and athletic outcomes.

• Current Limitations: Room assignments are often made based on administrative convenience, personality guesses, or limited preferences, ignoring critical biological and environmental factors like chronotypes (circadian rhythms) and sleep hygiene needs (noise, light, temperature).
• Consequence: These mismatches result in increased sleep latency, reduced total sleep time, and diminished daytime functioning. Existing solutions lack a scalable, data-driven framework to systematically align roommates based on sleep science.

2. Methodology

The authors propose a two-part solution: a standardized data collection instrument and a custom algorithmic matching tool.

A. The Instrument: ROOMPREF Questionnaire

A brief, eight-item survey designed to capture key sleep variables in under two minutes. It collects:

• Demographics: Sex (and preferred roommate sex).
• Chronotype: Categorized into Morning, Neutral, or Evening types based on self-reported preference.
• Environmental Sensitivities: Tolerance levels for noise, lighting, and temperature.
• Behavioral Factors: Snoring behavior (Yes/No).
• Optional: Open-text comments for specific roommate requests.

B. The Tool: ROOMPREF Web Application

A free, web-based clustering tool built using Python, pandas, scikit-learn, and Streamlit. The workflow involves:

1. Data Ingestion: Administrators upload CSV data exported from a Google Form version of the questionnaire.
2. Stratification: Participants are first grouped by Sex to ensure gender-appropriate clustering.
3. Chronotype Filtering: Participants are mapped into Morning, Neutral, or Evening categories. The algorithm enforces a constraint to avoid pairing "Morning" and "Evening" types in the same room unless absolutely necessary (flagged as a warning).
4. Conflict Exclusion Logic:
   • Snoring: Participants who snore are excluded from rooms with participants requiring "Absolute Silence" unless no other option exists (flagged).
5. Adaptive Clustering:
   • Algorithm: K-Means clustering is applied within sex–chronotype subgroups.
   • Heuristic for $k$: The number of clusters ($k$) is determined adaptively based on group size ($N$):
     • If $N \leq 3$: $k=1$ (single cluster).
     • If $N \geq 4$: $k=2$ (split into two clusters).
   • Features: Clustering is based on label-encoded and z-score standardized variables for Noise, Lighting, and Temperature preferences.
6. Post-Cluster Swapping Algorithm:
   • If a conflict remains (e.g., a snorer paired with a silence-sensitive person), the algorithm searches within the same cluster for a "tolerant" roommate.
   • It performs a round-robin swap to relocate the sensitive participant to a room with a tolerant roommate, ensuring room size constraints are met.
7. Output: The system generates a table of assignments, conflict flags, distribution charts, and downloadable CSV files. Users can toggle between manual review and automatic allocation.

3. Key Contributions

• Methodological Innovation: Introduces a structured, reproducible framework for roommate allocation that integrates sleep science (chronotypes and environmental preferences) into logistical decision-making.
• Open-Source Resource: Provides a free, accessible web tool and the underlying code (hosted on GitHub) for administrators, coaches, and residential staff, removing the need for specialized programming skills.
• Algorithmic Logic: Develops a hybrid approach combining K-Means clustering for environmental similarity with rule-based constraints for biological incompatibility (chronotypes/snoring), followed by a heuristic swapping mechanism to resolve residual conflicts.
• Scalability: Designed to be easily deployed across diverse sectors (sports, military, education) with minimal technical overhead.

4. Results (Pilot Study)

The tool was piloted with 51 Paralympic athletes (21 female, 30 male) ahead of the 2024 Paris Games.

• Workflow: The process successfully moved from survey distribution (Google Forms) to data export (CSV), upload to the app, and generation of roommate pairs.
• Visualization: The tool successfully generated distribution charts for chronotypes and environmental preferences, allowing coordinators to visualize the cohort's sleep profile.
• Allocation: The system produced a final list of room assignments with conflict flags (e.g., identifying potential snoring issues) and allowed for automatic allocation based on a target room size (2 per room).
• Outcome: The pilot demonstrated the feasibility of the end-to-end pipeline in a real-world, high-pressure environment, proving the tool can operate within existing organizational workflows.

5. Significance and Limitations

Significance:

• Practical Impact: Offers a pragmatic solution to a common logistical problem, potentially improving sleep quality and performance without the prohibitive cost of single-room accommodations.
• Evidence-Based: Moves roommate assignment from intuition to data-driven decision-making, explicitly addressing the known negative impacts of circadian misalignment and environmental disturbances.
• Versatility: Applicable to any shared living context, from university dorms to military barracks.

Limitations:

• Lack of Empirical Validation: The study describes the tool's development and workflow but does not yet provide empirical data on whether the matching actually improves sleep outcomes, wellbeing, or performance.
• Psychometric Validity: The questionnaire has not undergone rigorous psychometric testing (reliability, factor analysis).
• Self-Reporting Bias: Reliance on self-reported data (e.g., snoring) may lead to inaccuracies due to social desirability bias or lack of self-awareness.
• Scope: The algorithm focuses strictly on sleep variables and does not currently account for broader psychosocial factors like personality compatibility or cultural considerations.

Future Directions:
The authors call for randomized controlled trials to validate the tool's impact on sleep metrics (actigraphy, sleep diaries) and performance. Future iterations may include more complex clustering algorithms, additional preference dimensions, and integration into broader digital management ecosystems.