Today, information technology has become a necessity for successful business solutions. In business advance estimated information can be the key for success. Today's multinational companies, almost without exception have got a group, which work with different mathematical, statistical or data mining methods to effectively predict information regarding the future. Groups for business activity use this predicted information to maximize profit for businesses. At this point, business and information technology are connected.
In my thesis I will describe forecasting models for time series from energy industry with statistical and data mining methods. Primarily – originating from E.ON Hungary – examining electricity data I will use four methods for time series forecasting. These are linear regression, auto regression adapted to noise, the Box and Jenkins developed autoregressive and moving average (ARMA) models, as well as neural network used in data mining. I'll use IBM SPSS Statistics for statistical modeling and NeuralWorks Predict programs for data mining modeling.
In addition to electricity data I will also mention gas consumption and demonstrate in time series coming from European Energy Exchange (EEX), that time series forecasts can not only be used for electricity data, but that they play a significant role in the energy industry.
I will also examine the efficiency of the forecast, and with help of this, I will compare different models and their variants.
There is a program made for the prediction task that enables fast and effective forecasting, based on the IBM SPSS Statistics results.