Data Cleaning Using Batch Reinforcement Learning
- Ghalage Prajakta J.
- Sonawane Nalini N.
- Tamhane Madhuri S.
- Deshmukh Nutan Subhash
Missing data, Uncertain Data, Reinforcement Learning, Q-Learning, Discretization Algorithm, UCAIM, Poor Attribute Removal.
In real world raw data is highly affected by Missing value and uncertainty. This missing and uncertain data leads some distraction in dataset. So that before storing that data in dataset we have to clean that data first. Data cleaning is an important step in data mining [3]. In this paper we introduce some methods to find and remove the missing data and uncertainty. We generate the missing data using Q-Learning Algorithm. In Q-Learning Algorithm the missing data is generate and replaces the Null values with generated one. We use new Discretization Algorithm called UCAIM (Uncertain Class-Attribute Interdependency Maximization) that will find and replace uncertain data. Batch Reinforcement Learning is area of machine learning. By using batch reinforcement learning we can batch the transitions. Removal of Poor Attribute is also part of data mining, especially for high dimensional datasets. We use attribute selection algorithm for the selection and removal of poor attribute.
Ghalage Prajakta J., Sonawane Nalini N., Tamhane Madhuri S., Deshmukh Nutan Subhash. "Data Cleaning Using Batch Reinforcement Learning".INTERNATIONAL JOURNAL OF ENGINEERING DEVELOPMENT AND RESEARCH ISSN:2321-9939, Vol.2, Issue 1, pp.610-613, URL :https://rjwave.org/ijedr/papers/IJEDR1401109.pdf
Volume 2 Issue 1
Pages. 610-613
