A new prediction model to discover probability of document in the website

Vocational teachers and students in Taiwan nearly have the amount of people 1,200,000. The Website of The Technological and Vocational Education is Skill duty teachers and students and the expert's knowledge entry website. This research picks user's data of the route to collect as the training data that originally study the way from the Data Warehouse of websites at first. Analyze the user's possible behavior route and describe models with set up its dynamic action with Petri Nets. Naive Bayesian Classification in Data Mining calculates that trains data to concentrate probability value of the network user's behavior route attribute, as predicting whether the next user looks for to the file. By the discovery rate of the file, put forward the amendment and problem and solve the mechanism to the webpage system or the structure. Find the following result while studying: 1. If data is it for independence each other can spend less and every attribute to link because every attribute connect with influence it of product fruit value 0. In order to overcome this question, originally research and propose and adopt M-estimate to revise the mechanism, can really improve this question through verifying. 2. The research random access 150 data in database of website. Use Naive Bayesian Classification to ask out every attribute probability, and the construction Data Mining engine and random access 100 testing data. Among them find, use shellfish's probability classification, find the file and predict that success rate is all more than 93%. 3. The button name of Website of The Technological and Vocational Education is not a clear definition, the ones that let users not know the file are preserved to the place, and some sub browsing through rate of webpage is too low, this research also proposes proper revision and recommendations.