title: SARIMAX Model for Energy Consumption Prediction
description: >-
This model predicts energy consumption based on meteorological data and
historical usage.
license: gpl
SARIMAX Model for Energy Consumption Prediction
Description
This SARIMAX model predicts energy consumption over a 48-hour period based on meteorological data and historical energy usage from 2021 to 2023. It utilizes time series data from a transformer station and incorporates both time-based and weather-related variables to enhance prediction accuracy.
Model Details
Model Type: SARIMAX (Seasonal ARIMA with exogenous variables)
Data Period: 2021-2023
Variables Used:
Lastgang: Energy consumption dataStundenwertStrahlung: Hourly radiationGlobalstrahlung_15Min: Global radiation every 15 minutesStrahlungGeneigteFläche: Radiation on inclined surfacesTheorPVProd: Theoretical photovoltaic productionDirektnormalstrahlung: Direct normal radiationSchönwetterstrahlung: Clear sky radiationLufttemperatur: Air temperatureLastgang_Moving_Average: Moving average of energy consumptionLastgang_First_Difference: First difference of energy consumption
Features
The model splits the data into training and testing sets, with the last 192 data points (equivalent to 48 hours at 15-minute intervals) designated as the test dataset. It defines target variables (Lastgang) and explanatory variables including hourly and daily patterns as well as derived features from the consumption data. The dataset includes preprocessed features such as scaled energy consumption (Lastgang), and weather-related features.
Installation and Execution
To run this model, you need Python along with the following libraries:
pandasnumpymatplotlibstatsmodelsscikit-learnpmdarimapsutil
Steps to Execute the Model:
Install Required Packages conda install pandas numpy matplotlib scikit-learn statsmodels pmdarima psutil or pip install pandas numpy matplotlib scikit-learn statsmodels pmdarima psutil
Load your Data
Preprocess the data according to the specifications
Run the Script