Abstract:Based on data frommanufacturer and method of empirical analysis, this paper address problem ofouter package damage and try to use rough set theory to trace reason of causingdamage. After attribute reduction of decision table, 14 rules are generated,which try to show inherent law of damage rate improvement. In the end of thepaper, the authors build a warning model of artificial neural network to remindrelative manufacturers to take measure to solve problem in advanced. 

Ⅰ. Introduction

In transportation of logistics,it is a common phenomenon that damaged outer package of merchandise Normally wecan see damages including common dent, water stained, collapse, holes and soon. All of these will bring possible loss or damage of merchandise, and evennegative impact to goods delivery and business reputation, which often make relativemanufacturers in trouble.  

There are many reasons causing box damage, such as quality of carton;the strength of the carton designed, skill level of workers who operating forkliftat production line or in warehouse, goods piled up in order, frequency of transportingtransition, rain, humidity, loading and unloading, workload of operator and soon. Take an example, collapse of cartoncornermay is caused not only by non-standardconveying or piling up but also associated with surrounding humidity or rain oreven a variety of reasons. Soit is difficult for merchants to judge rootcause of damage and work out a good solution to improve work-flow related tocut down the damage rate of the packageeffectively. Therefore it is necessary to target root reason by analysis andidentification of damage, findhindering reasons and reveal lawofpackage being broken. Just by which thoseproblems maybe resolved efficiently. 

Rough Set Theory isundoubtedly a powerful analatical tool of deeply reason mining.The theory was raised by matemaPawlak Z of Polandin early 80’ of last century andquickly gained a wide acceptance and development because of successfulapplication in many fields^[1]. It is a new math tool to deal withquestions of vague and non-precision by data analysis, classification andreasoning to the relationship between data, which will help to find implicitknowledge and reveal the potential law and provide support to decision making^[2].Since 1990, the rough set theory has gotten a wide concern because of itsunique advantage in incomplete data and imprecise knowledge analysis. Inparticular, it’s function of knowledge reduction can classify informationsystems by reduce attribute at the same discernbility capacity and generaterule of classification and decision-making^[3]. That is reason whyrough set theory has become one of the most effective way in data mining.