Energy-Efficiency-Project

Description: The energy efficiency dataset analysis is done using 12 different building shapes simulated in Ecotect.
Software: Python 3.7
Presentation: Jupyter Notebook
Goal: Predicton of Heating and Cooling Loads for various buildings

Tasks

1. Load libraries
2. Data Cleaning and Transformation
3. Modeling
        3.1 Regression
        3.2 Classification
   

Details on Regression and Classification

3.1 Regression

Targets: Two kinds; heating load, and cooling load
Explanatory Variables: relative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distribution

Models

1. Linear Regression (Single Output and Multiple Output)
2. Lasso Regression
3. Ridge Regression
4. Polynomial Regression
5. KNN Regressor
6. Support Vector Regressor
7. Random Forest Regressor
8. Gradient Boosting Regressor (finally chose this)

Best train accuracy ~ 99.8% and Best validation accuracy ~ 99.2%

3.2 Classification

Model: Artificial Neural Network (One input layer, one hidden layer, and one output layer)
Classes: 3 categories for both loads, broken down by looking at histograms
Parameter tuning: Using grid search
Package: KERAS

Tuned parameters

1. Epochs (200)
2. Batch size (20)
3. Optimizer (Nesterov Adam optimizer)
4. Learning rate of the optimizer (0.1)
5. Initialization mode of weights (normal)
6. Activation function (softsign)
7. Drop out rate and weight constraint (0.3, 3)
8. Number of neurons in the hidden layer (5)

Best train accuracy ~ 96% and Best validation accuracy ~ 94%