Abstract: Intra-day periodicity has been widely found in financial high frequency data study,It is a dynamic effect characterized by intra-day periodic motion and it affects the accuracy of econometric model estimation which contains intra-day financial variables. The importance of intra-day periodic adjustment is discussed firstly in this study and then introduces self-organizing maps as a intra-day periodic adjustment solution are introduced based on financial ultra high frequency duration data. The SOM method is a feature extraction on the basis of neural network learning which can recognize the dynamic feature in high-dimensional data in order to overcome the disadvantage of static periodic adjustment. Finally a monte carlo simulation through autoregressive conditional duration model is built to compare the effects of three intra-day periodic adjustment methods. The result shows that the SOM method performs more effective and stable.Therefore SOM method can be particularly suited for analysis of periodic structure in big data.

Key words: intra-day periodicity, autoregressive conditional duration model, SOM, periodic adjustment