Search

In this paper we perceive data analysis with empirical probability functions as a data mining method. We propose a way to carry out this type of analysis by employing the LISp-Miner system, namely the CF-Miner procedure and pattern difference quantifiers. In order to confirm that LISp-Miner is a suitable tool for this purpose, we briefly present both methods and then show their equivalence. We do this by providing theoretical description which we then support by analysing a small set of data concerning traffic accidents with methods and comparing results. Afterwards we provide an example of analysis of a full data set concerning rail tickets sold at selected stations in 2014. We show that by considering “difference histograms” it is possible to identify remarkable dissimilarities in histograms of time of ticket sale that would not be found otherwise. Both analyses confirms that the method we propose can provide new and interesting results even if the data has been already analysed.

This article deals with the use of data mining within the area of exploratory data analysis in fare collection systems. It describes the fare collection system and its model defined for the needs of making exploratory data analysis. The model contains inputs and outputs, internal functions and random impacts. The exploratory data analysis has been carried out both in spreadsheet and through the data mining system LISp-Miner. The conclusion of the paper outlines the possibilities of further extension and use of the GUHA procedure in the LISp-Miner system.