A data mining algorithm based on joint distribution rules in disaster risk valuation

摘要

A jointing distribution data was mined by copula algorithm for disaster risk control in this paper. We illustrate all these concepts with an example of severe dust storm and consider samples of simultaneous observations in most important two numerical hazard variables of maximum wind speed and duration. The analysis emphasizes how inappropriate in the variables jointing can become in cases of significant departure from the linear hypothesis. The results show that with the help of the cumulative probability curve according to put forward a kind of cleaning association rules based on the copula algorithm, the multiple variables jointing can be engaged and the differences of goodness fit between copula jointing data and linear jointing data can be visually compared. Based on the optimization of the evaluation parameters in copulas model, bivariate jointing distribution can be completely captured by the Frank copula model by demonstrating its usefulness and efficiency from the data tracking of accuracy for dust storms disaster.