Forecasting Electricity Market Price

1
Forecasting Electricity Market Price Spike using
Data Mining Techniques
Presenter Liew Eng Ru (40167918)
Supervisor Dr Zhao Yang Dong
Introduction Volatility in the energy markets
makes it an imperative to understand potential
scenarios for future spot prices to facilitate
informed decision making. Price spike forecasts
are a fundamental input to an energy companys
decision-making and strategy development. Profits
will be directly affected by the accuracy of
these forecasts. Aim This thesis focuses on
price spike analysis using data mining techniques
based on historical data extracted from the ESAA.
The probability of the occurrence of price spike
in the four different seasons will be calculated
based on the historical data mined. And a very
simple forecast model will be proposed to predict
the increase in electricity price based on the
increase in average electricity demand.
Background Price Spike is the sudden jumps in
energy prices, it can be hundreds of times higher
than the normal electricity market energy price
(e.g. 6000 vs.. 30/MWh). Price Spike is often
driven by unexpected events such as a poor
weather forecast, unusual demand patterns
etc.
Data Mining "Data mining is defined as the
process of discovering patterns in data. It is
about finding and describing patterns in data
which will lead to some advantage (usually
economic advantage). Historical Data
Date, daily average electricity demand
and market price
Next
Next
Data Mining
Data are categorized according to its price range
Data are group according to the low and high
demand being set
Weeks are clustered base on the four seasons
Data Mining results
Probability Calculation Using the Data mining
results, probabilities are calculated
Forecast Model - The Regression Model Regression
analysis is concerned with modeling the
relationships among variables. If the true
relationships among the variables were known
exactly, the investigator would be able to
understand, predict, and control the response.
The true relationships among the studied
variables, however, will rarely be known, and one
must rely on empirical evidence to develop
approximation.
References 1. Zhao Yang Dong "A Neural Network
Based Framework for Electricity Load and Price
Forecasting", Proc. Asian Control Conference
2002. 2. Xin Lu, Zhao Yang Dong and Xue Li,
"Electricity Market Price Spike Forecast with
Data Mining Techniques", to appear in
International Journal of Electric Power Systems
Research. 3. Prabhas Chongstitvatanna "Data
Mining an Overview", Lecture notes to Master
degree students of the computer science, faculty
of science, Kasetsart