/Sensor_Fusion_AutoML

Fusion of data from Multiple sensors to detect fire

Primary LanguagePython

Sensor_Fusion_AutoML

Coursework Project for Applied Machine Learning for Mechanical Engineers

Objective

Fusion of data from Multiple sensors to detect fire using AutoML

Dataset

The dataset consists of 7 sensors and timestamps. The sensors are:

  1. Temperature
  2. Humidity
  3. Pressure
  4. VOC
  5. Carbon Di Oxide
  6. Hydrogen and Ethanol concentration
  7. Particulate Matter concentration

AutoML Overview

AutoML is a process of automating the end-to-end machine learning process. It includes:

  1. Data preprocessing
  2. Feature engineering
  3. Model selection
  4. Hyperparameter tuning
  5. Model deployment

Libraries Used

  1. FLAML
  2. MLJAR

Results

FLAML Analytics

Best Learner LGBM (Decision Tree)
Time for finding Best Model 5.19 Seconds
Best Model Accuracy 99.98%
Training Time 0.053 Seconds
Final Learning Rate 0.09

MLJAR Analytics

Best Model Type Ensemble
Models Used Xgboost(4) and Neural Network(1)
Accuracy 99.98%
Training Time 1.83 Secs