Prediction and Analysis for the Broken-down Number Per Thousand Cars
Xiaolian Zheng · Shuxue de shijian yu renshi · 2005
Responding to the drawbacks of the old statistical method of the number of down-broken cars per thousand, this paper attempts to improve the existing statistical method to recreate a new one, namely we redefine the number of down-broken cars per thousand and utilize the cluster analysis in data mining to establish a universal operational model, so as to comprehensively consider those groups which members share certain properties in common.For the questions of scanty of data and short-term forecast,we adjust the universal operational model and study the different weight among the same class.Base on the weighted data, we can get the value of forecast through fitting the curve. Since acute shortage of data in long-term forecast, we start from the view point of pure statistics and get the forecast value. The forecast model can be used comprehensively. This paper solves the model mentioned above by using SAS and MATLAB, the forecasted result with high veracity of and conforms to practical condition.