Time Series Analysis of Energy consumption pattern using XGBoost and other Machine Learning models
Steps done include the following: 4.1 Dataset Loading and Preparation: 31 4.1.1 Loading Energy and Weather Dataset: 31 4.1.2 Converting time features to datetime index: 32 4.2 Data Cleaning and Analysis 33 4.2.1 Energy Dataset Cleaning 33 4.2.2 Energy Data Analysis 35 4.2.3 Weather Dataset Cleaning 38 4.2.4 Weather Data Analysis 44 4.3 Dataset Integration 45 4.3.1 Merging Energy and Weather Dataset 45 4.4 Target Feature Analysis 50 4.4.1 Analyzing the Target Feature in the Dataset 50 4.5 XGBoost Model Development with Temporal Features 53 4.5.1 Train-Test Split 53 4.5.2 Creating Temporal Features 54 4.5.3 Temporal Feature and Target relation 55 4.5.4 Model Training and Prediction 57 4.5.5 Evaluation Metrics 59 4.6 XGBoost Model Development with Temporal and Lag Features 59 4.6.1 Creating Lag Features 59 4.6.2 Time Series Cross-Validation 60 4.6.3 Model Training with Temporal and Lag Features 61 4.6.4 Evaluation Metrics 63 4.7 XGBoost Model Development with Temporal, Lag and Additional Features 64 4.7.1 Creating Additional Features 64 4.7.2 Model Training with Temporal, Lag and Additional Features 64 4.7.3 Evaluation Metrics 65 4.8 Hyperparameter Tuning 66 4.9 Support Vector Regression Model Development 68 4.10 Random Forest Model Development 70 4.11 Gradient Boosting Model Development 72 5 Results and Visualization 74 5.1 Analysis and Visualization of XGBoost Model Performance 74 5.2 Analysis and Visualization of XGBoost Hyperparameter Tuning 76 5.3 Analysis and Visualization of Feature Importance 78 5.4 Comparison of XGBoost Performance with other models 80