obinopaul/Predicting_hourly_energy_consumption_using_Prophet

This repository contains Python code and documentation for univariate time series modeling on PJMs Eastern Interconnection Grid, using Prophet. The aim is to forecast energy consumption based on hourly data spanning 10 years. The results achieved good accuracy, with forecasted values available for future hours.

Hourly Energy Consumption from PJM's Eastern Interconnection Grid

==============================

This repository contains code for univariate time series modeling of hourly energy consumption data (in Megawatts) from PJMs Eastern Interconnection Grid. The aim of this project is to build a model that can accurately forecast hourly energy consumption (in Megawatts) based on historical data.

Dataset

The data used in this project is a time series dataset containing hourly energy consumption data from PJMs Eastern Interconnection Grid, spanning over 10 years. The dataset is available in the dataset folder of this repository.

Methodology

The analysis was performed using Facebook's Prophet library in Python. Prophet is a powerful tool for time series forecasting that accounts for seasonality, trends, and holidays.

The following steps were followed to perform the modelling:

1. Data preprocessing: cleaning, formatting, and resampling the data to hourly frequency
2. Exploratory data analysis: The data was visualized to gain insights into the patterns and trends in the data.
3. Model fitting: training the Prophet model on the preprocessed data
4. Model validation: evaluating the accuracy of the model using metrics such as Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE)
5. Forecasting: generating future predictions based on the fitted model

Result

The model achieved good accuracy on the validation data, with an MAE of 190 and an RMSE of 249. The forecasted values for the next Z hours are available in the "models" folder of this repository.

Usage

Clone this repository to your local machine.

git clone https://github.com/obinopaul/hourly_energy_consumption.git                                       

Install the required packages.

pip install -r requirements.txt                         

Run the notebook.
Run the hourly_energy_consumption_2.ipynb notebook to preprocess the data, fit the model, and generate predictions

Conclusion

In this project, we successfully built a univariate time series model using Prophet to forecast hourly energy consumption in PJMs Eastern Interconnection Grid. The model achieved good accuracy on the validation data, and the predictions for the next several hours are available for use. "# hourly_energy_consumption"