Heat Measuring and Prediction Demo By Akash Pandey

This notebook provides a comprehensive demonstration of generating synthetic time-series data related to environmental factors and building energy, followed by various data analysis, visualization, and machine learning modeling techniques.

Table of Contents

1. Overview
2. Features
3. Synthetic Data Generation
4. Plotting Utilities
5. Modeling - Heating Prediction
6. Modeling - Air Quality Prediction (PM2.5)
7. How to Run the Demo
8. Number System Converter

Overview

This project simulates environmental and building data (outside temperature, humidity, heating load, and PM2.5 air quality) and demonstrates a workflow for:

• Generating realistic synthetic time-series data.
• Visualizing time-series, correlations, and model predictions.
• Building simple machine learning models (RandomForestRegressor) to predict heating load and PM2.5 levels.
• Analyzing data under different scenarios (default, cold wave, pollution spike).

Features

• Synthetic Data Generator: Creates hourly data for outside temperature, humidity, heating demand, and PM2.5 air quality with seasonal cycles, trends, and random noise/spikes.
• Plotting Utilities: Functions to visualize time series, scatter plots with linear fits, correlation heatmaps, and forecast vs. actual comparisons.
• Machine Learning Models: Includes functions to prepare features, train, and evaluate Random Forest Regressor models for predicting heating demand and PM2.5 levels.
• Scenario-based Demos: Demonstrates data generation and analysis under 'default', 'cold_wave', and 'pollution_spike' scenarios.

Synthetic Data Generation

The generate_synthetic_dataset function is the core of the data simulation. It allows you to create data for a specified period and frequency, with options to introduce different scenarios affecting temperature or air quality.

Key Functions:

• generate_time_index(start_date, periods, freq): Creates a pandas DatetimeIndex.
• seasonal_signal(...): Generates a signal with daily and yearly seasonality, trend, and optional spikes.
• generate_synthetic_dataset(...): Orchestrates the generation of all features based on specified parameters and scenarios.

Plotting Utilities

Several functions are provided for quick visualization of the generated data and model results.

Key Functions:

• plot_time_series(df, cols, title, figsize, savepath): Plots selected columns of a DataFrame over time.
• plot_scatter_with_fit(x, y, xlabel, ylabel, title, savepath): Creates a scatter plot with a linear regression fit.
• plot_correlation_heatmap(df, cols, title, savepath): Displays a correlation matrix heatmap for specified columns.
• plot_forecast_vs_actual(dates, actual, predicted, title, savepath): Compares actual values against predicted values over time.

Modeling - Heating Prediction

This section focuses on predicting the building's heating load (heating_kW) based on environmental factors and lagged features.

Key Functions:

• prepare_features(df, target_col): Selects features and target, handling NaNs.
• train_evaluate_regressor(X_train, y_train, X_test, y_test, model): Trains a regression model (defaulting to RandomForestRegressor) and evaluates its performance (MSE, R2).

Modeling - Air Quality Prediction (PM2.5)

This section demonstrates predicting PM2.5 air quality using a similar machine learning approach, incorporating feature scaling.

Key Functions:

• train_simple_pm25_model(df): Prepares features, scales them, trains a RandomForestRegressor for PM2.5, and returns performance metrics.

How to Run the Demo

The demo_all() function encapsulates the entire workflow for a given scenario.

1. Execute the first code cell (0FTiMlKYEX-_) to define all functions and classes.
2. The if __name__ == '__main__': block at the end of the first code cell will automatically run the demo_all() function for three different scenarios: 'default', 'cold_wave', and 'pollution_spike'.
3. Observe the printed output and generated plots for each scenario.
4. You can modify the demo_all() function or the if __name__ == '__main__': block to experiment with different parameters or scenarios.

Number System Converter

Separately, the notebook includes a simple command-line interface (CLI) for converting numbers between different bases (decimal, binary, octal, hexadecimal).

Key Functions (in cell qk34UxAnJ0zX):

• decimal_to_binary, decimal_to_octal, decimal_to_hexadecimal
• binary_to_decimal, binary_to_octal, binary_to_hexadecimal
• octal_to_decimal, octal_to_binary, octal_to_hexadecimal
• hexadecimal_to_decimal, hexadecimal_to_binary, hexadecimal_to_octal

To use the converter:

1. Execute the code cell (qk34UxAnJ0zX) containing the converter functions and the main loop.
2. Follow the prompts in the output to select a conversion type and enter a number.
3. Choose option 13 to exit the converter.

Working by Akash Pandey for Learning Purpose only

give realtime data in csv for more accurate output