Correlation Mining between Time Series Stream and Event Stream

There are huge amounts of temporal data in many real-world applications. National climate databases provide examples of this type of temporal data. Two categories of data are held in these databases: the first are time series events such as relative humidity, amount of cloud, visibility, soil temperature, bulb, wind direction, and so forth which are recorded in hourly-bin time intervals, and the second are event steam data like heavy rains, flooding, hurricane, ice storms and so forth. In this work we propose a general framework to explore the correlation between time series streams and events stream using equation discovery method. 4 well designed time series system measures a bunch of parameters while many event may happen independently from time series interval, it means that an event has its own time point. We try to find the effective parameters on events occurrences. Firstly, we find the closest time series record for any events; therefore, we have gathered different parameters value when an event is occurring. Using LA GRANGE's equation discovery method we find the equation between the events and effective parameters. LAGRANGE's equation discovery is one of the well-known equation discovery methods which analyze a great domain of equations. This equation and time series model can predict future events efficiently. Our data set is Tehran daily aerological data set which is digital data set archived at the Iran meteorological organization. This data set has statistics for geopotential height, temperature, specific humidity, Zonal wind, and meridional wind Data has collected from Tehran "Mehrabad" station. The earliest data is from 1961; the latest, from 2005 (44 years). This data set includes 17 types of events. Time series models can predict next time series parameters value and by using these equations the closest event can be predicted. The LAGRANGIAN method is able to find some suite equation with R>0.98 and s<0.008, which R is the correlation coefficient and S is the least-square. This work is the first estimate in the area of correlation mining for a huge data set of aerology and can be extended in many different data sets in any other environments.