Smart Home Satisfaction Prediction with Random Forest and Real-Time Simulation

This project explores how machine learning and adaptive systems can be applied to optimize environmental conditions in a smart home environment. The core objective is to predict occupant satisfaction based on environmental data and simulate real-time adjustments to improve comfort. This is achieved using a Random Forest model and a real-time simulation system.

Project Overview

This project focuses on leveraging environmental data, such as CO2 levels, indoor temperature, and humidity, to predict how satisfied occupants are in a smart home. The system uses Random Forest regression to estimate satisfaction and then applies adjustments like heating or ventilation to optimize comfort based on these predictions.

Breakdown of the steps:

• Exploratory Data Analysis (EDA):
  Data is first cleaned and explored to uncover key relationships between environmental factors and occupant satisfaction.

• Predictive Model:
  A Random Forest Regressor is used to predict satisfaction based on factors like indoor temperature and humidity. The model is saved and used for simulation purposes.

• System Simulation:
  A real-time simulation applies control logic to adjust environmental conditions based on the predicted satisfaction. The simulation dynamically responds to environmental changes to maintain comfort.

Results

The Random Forest model successfully predicts satisfaction based on environmental data. It is used in the simulation to guide real-time adjustments in a smart home, such as:

• Heating adjustments to increase temperature if satisfaction levels drop.
• Ventilation based on CO2 levels to maintain optimal air quality.

The system continuously monitors and optimizes environmental conditions, demonstrating how adaptive control systems can improve occupant comfort.

Predictive Model and Scaling

The Random Forest model and MinMaxScaler are trained on environmental data to predict satisfaction levels. These models are saved as random_forest_model.pkl and scaler.pkl, respectively, for reuse in the simulation.

Execution Overview

• Exploratory Data Analysis: Run exploratoryDA.ipynb to clean, explore, and visualize the data.
• Satisfaction Prediction: Execute intelligence.ipynb to build and save the Random Forest model.
• Real-Time Simulation: Use system.ipynb to simulate the smart home environment. You can also run simulation.py for a live visual interface.

Potential Future Improvements

• Incorporating more environmental factors such as lighting or noise levels.
• Expanding the system to include energy-efficient decision-making, optimizing not only comfort but also resource consumption.

---

---

Have a great day :)