Abstract:This paper presents a new hybrid model for predicting German electricity prices. The algorithm is based on combining Gaussian Process Regression (GPR) and Support Vector Regression (SVR). While GPR is a competent model for learning the stochastic pattern within the data and interpolation, its performance for out-of-sample data is not very promising. By choosing a suitable data-dependent covariance function, we can enhance the performance of GPR for the tested German hourly power prices. However, since the out-of-sample prediction depends on the training data, the prediction is vulnerable to noise and outliers. To overcome this issue, a separate prediction is made using SVR, which applies margin-based optimization, having an advantage in dealing with non-linear processes and outliers, since only certain necessary points (support vectors) in the training data are responsible for regression. Both individual predictions are later combined using the performance-based weight assignment method. A test on historic German power prices shows that this approach outperforms its chosen benchmarks such as the autoregressive exogenous model, the naive approach, as well as the long short-term memory approach of prediction.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem of price prediction in the German electricity market. Specifically, the author proposes a new hybrid model that combines Gaussian Process Regression (GPR) and Support Vector Regression (SVR) to improve the prediction accuracy of German electricity prices. #### Background and Challenges 1. **Market Characteristics**: - The electricity market is different from other energy commodities (such as oil or natural gas), being non - storable and weather - dependent. - With the increase in the proportion of renewable energy in total energy production, the volatility of the electricity market has increased significantly, especially due to the intermittent changes in weather conditions. - Factors such as non - stationarity of data, complex correlations, and time - correlations make electricity price prediction more complex. 2. **Limitations of Existing Methods**: - Traditional methods such as Autoregressive Exogenous Model (ARX) and Long - Short - Term Memory Network (LSTM) perform poorly in handling complex and non - linear relationships. - A single model (such as GPR or SVR) has deficiencies in handling noise and outliers, resulting in less accurate prediction results. #### Proposed Solutions To overcome the above challenges, the author proposes a new hybrid model based on the combination of multi - kernel Gaussian Process Regression and kernel - based Support Vector Regression. The specific improvements are as follows: 1. **Gaussian Process Regression (GPR)**: - GPR is good at learning random patterns in data and performing interpolation, but its prediction performance for out - of - sample data is not ideal. - By selecting an appropriate data - dependent covariance function, the prediction performance of GPR on test data can be enhanced. - However, GPR's out - of - sample prediction depends on training data and is easily affected by noise and outliers. 2. **Support Vector Regression (SVR)**: - SVR adopts a margin - based optimization method and has an advantage in handling non - linear processes and outliers. - Only certain necessary points (support vectors) in the training data are responsible for regression, so it is more robust to noise and outliers. 3. **Hybrid Model**: - The individual prediction results of GPR and SVR are combined through a performance - based weight - allocation method to form the final prediction result. - This hybrid model can fully utilize the probability framework of GPR and the non - linear processing ability of SVR, thereby improving the accuracy and robustness of prediction. #### Experimental Verification Through the test of historical German electricity price data, the results show that this hybrid model is superior to selected benchmark models, such as ARX, the naive method, and LSTM. ### Summary This paper proposes a new hybrid model by combining the advantages of GPR and SVR, aiming to improve the accuracy and robustness of price prediction in the German electricity market, especially performing well in the face of complex market fluctuations and data characteristics.

Electricity Price Prediction Using Multi-Kernel Gaussian Process Regression combined with Kernel-Based Support Vector Regression

Predicting long-term electricity prices using modified support vector regression method

Predictive Accuracy of a Hybrid Generalized Long Memory Model for Short Term Electricity Price Forecasting

A Hybrid Approach for Probabilistic Forecasting of Electricity Price.

Short-term Electricity Price Forecasting with Empirical Mode Decomposition Based Ensemble Kernel Machines.

Electricity Demand Forecasting with Hybrid Statistical and Machine Learning Algorithms: Case Study of Ukraine

Probabilistic mid- and long-term electricity price forecasting

Interpretable Interval Prediction-Based Outlier-Adaptive Day-Ahead Electricity Price Forecasting Involving Cross-Market Features

Data-driven Two-step Day-ahead Electricity Price Forecasting Considering Price Spikes

Price forecasting in the Ontario electricity market via TriConvGRU hybrid model: Univariate vs. multivariate frameworks

Short- and long-term forecasting of electricity prices using embedding of calendar information in neural networks

Electricity price forecasting using sale and purchase curves: The X-Model

Day-ahead electricity price prediction applying hybrid models of LSTM-based deep learning methods and feature selection algorithms under consideration of market coupling

Comparison of Classical and Nonlinear Models for Short-Term Electricity Price Prediction

An optimized deep learning approach for forecasting day-ahead electricity prices

Multivariate Probabilistic Forecasting of Intraday Electricity Prices using Normalizing Flows

Electricity Price Prediction using Machine Learning

Mixed Kernel Function Support Vector Regression with Genetic Algorithm for Forecasting Dissolved Gas Content in Power Transformers

Electricity Price Forecasting Model based on Gated Recurrent Units

Forecasting electricity prices with machine learning: Predictor sensitivity

Probabilistic forecasting of electricity consumption, photovoltaic power generation and net demand of an individual building using Gaussian Processes