Room Occupancy Prediction (PySpark)

A Big Data–driven machine learning project for predicting room occupancy using environmental sensor data (temperature, humidity, light, CO₂).

Built using PySpark to demonstrate scalable data processing, distributed computation, and real-world smart building applications.

Features

• Data processing with PySpark
• Exploratory Data Analysis (EDA)
• Machine learning models:
  • Logistic Regression
  • Decision Tree
  • Random Forest
• Model evaluation (Accuracy, F1 Score)
• Big Data scaling experiments
• Simulated large-scale dataset processing

Project Structure

├── data/ # raw and processed datasets
├── notebooks/ # exploratory and modeling notebooks
├── src/ # reusable project code
├── models/ # saved trained models
├── reports/ # figures and outputs
├── scripts/ # runnable pipeline scripts
├── README.md
├── requirements.txt
└── .gitignore

Setup

1. Open any of the notebooks in Google Colab

2. Generate GitHub Token

Go to https://github.com/settings/personal-access-tokens Create a new fine grained token

• Default settings should be fine Save off the token for later

3. Fetch your Google Drive path to the Google Drive folder

Should be something like "Spring 2026/CIS 5570/Big-data-group-1/Room-Occupancy-Detection/"

4. Define your Google Colab secrets

BIG_DATA_TOKEN {GitHub Token} GIT_NAME {Your Full Name} GIT_EMAIL {Your Email} BIG_DATA_PATH /content/drive/MyDrive + {Your Google Drive path to the shared drive}

5. If you swap notebooks you will need to re-enable these secrets

Dataset

UCI Occupancy Detection Dataset: https://archive-beta.ics.uci.edu/dataset/357/occupancy+detection

Goal

Predict whether a room is occupied using sensor readings and demonstrate how distributed systems like Spark can scale to real-world smart building scenarios.

Team

• Andrew Luwaga
• Prince Kwarteng Amaning
• Spencer Novaco
• Israel Sanchez