Decision Tree Framework and Relative Data Distribution Comparison of Markov Model for Privacy-Preserving Data Mining

Project Discription

Decision Tree Framework and Relative Data Distribution Comparison of Markov Model for Privacy-Preserving Data Mining

Abstract

The period of huge database is currently a major issue. Be that as it may, the conventional information examination will most likely be unable to deal with such extensive amounts of information. So analysts attempt to build up an elite stage to productively investigate and keep up the calculations. Here proposed work has resolve this issue of computerized information security by finding the connection between the columns of the dataset which depends on the markov model. So columns are stored on different servers which decrease overall dataset maintaince cost while security get increased. In other words whole work is broadly divided into few blocks such as pre-processing, frequent rule generation from the dataset which help in identifying the information, next is to filter those rules from the rule set which are frequent from others. Finally all filtered rules are grouped one by one so that all the relative attribute values will make minimum space requirement for data storage. So columns as per there markov bonding with other column is distribute on the different site. Here it was try to put strongly connected column on single site but due to limitations of the site storage, low bonded column is distribute to other site. So depend upon the relation between the columns data partition is done. By the utilization of Advanced  Encryption System calculation security of the information at server side get upgrade too. Analysis is done on genuine dataset. Results demonstrates that proposed work is better as contrast with different past methodologies on the premise of assessment parameters. As row insertion time is less while executing proposed work calculation. It has seen that by increment in dataset size, execution time also increments. As markov model has generate patterns of columns which reduce dataset size and execution time of row insertion is directly reduced.

Dependent Column

Now as per the different frequent rules of the dataset the relation between columns can be evaluate.

Here all possible pair of columns are prepared then find number of rules between them. So if total of rules present in the dataset column act as the bond strength between the columns. Sort Highly related Rules in other words pattern having highest number of rules in there group of columns is consider as the strongly related column group.

Column Distribution Structure

Here in this step whole columns as per there bonding with other column is distribute on the different site. Here it was try to put strongly column on single site but due to limitations of the site storage, low bonded column is distribute to other site. So depend upon the relation between the columns data partition is done.

Matlab Projects With Source Code

Project Sample Image