Data Mining and Knowledge Engineering

D. Minnie ¹, S. Srinivasan ²

Affiliations
1 Department of Computer Science, Madras Christian College, Chennai-600059, Tamil Nadu, IN
2 Department of Computer Science and Engineering, Anna University Regional Centre Madurai, Madurai, Tamil Nadu, IN

Abstract

Clinical Pathology uses laboratory tests on body fluids such as blood and urine to diagnose diseases. Haematology is the study of blood and blood forming organs such as bone marrow. In this paper we analyze the components of the bone marrow and the structure of the bone marrow analysis database. The Knowledge Discovery in Databases (KDD) steps are briefly explained. 18,000 bone marrow analysis records are collected from a reputed Hospital and this raw data is transformed into a preprocessed data using the pre-processing phases of KDD such as Data Cleaning, Data Selection and Data Transformation. The eliminate_the_tuple technique is used to clean the data. The attributes related to the bone marrow components are selected. The ranges of low, high and normal values for the individual attributes are used to transform the data. The Data Mining techniques are studied and theapriori algorithm is selected for finding frequent itemsets that are used for the generation of association rules. The transformed bone marrow data with low values is used to generate associations between the attributes of the bone marrow dataset.

Keywords

Association Rule Mining, Bone Marrow Analysis, Haematology, Knowledge Discovery in Databases.

Full Text

     

G. Manikandan ¹, S. Srinivasan ²

Affiliations
1 Sathyabama University, Chennai, IN
2 CSE Department, Anna University of Technology, Madurai, IN

Abstract

In this paper, we have presented a novel approaches for effectively mining of spatially co-located moving objects from the spatial databases. We propose a novel technique for co-location pattern mining which materializes spatial neighbor relationships with no loss of co-location instances and reduces the computational cost with the aid of the Prim's Algorithm. The spatially co-location mining technique is efficient since it generates and filters the candidate instances. Subsequently, the neighborhood relationships are carried out by the designed neighborhood and the node membership functions which satisfy the minimum conditional threshold. This paper has been inspired by the Join-less approach for mining spatial co-location patterns. We use a spatial database that contains the moving objects and its corresponding spatial location for spatial co-location pattern mining to mine spatially co-located moving objects.

Keywords

Spatial Data Mining, Co-Location, Prim's Algorithm, Moving Objects.

Full Text

     

G. Manikandan ¹, S. Srinivasan ²

Affiliations
1 Sathyabama University, Chennai, IN
2 Department of Computer Science, Anna University of Technology, Madurai, IN

Abstract

In this paper, we have presented a novel approaches for effectively mining of spatially co-located moving objects from the spatial databases. We propose a novel technique for co-location pattern mining which materializes spatial neighbor relationships with no loss of co-location instances and reduces the computational cost with the aid of the Prim's Algorithm. The spatially co-location mining technique is efficient since it generates and filters the candidate instances. Subsequently, the neighborhood relationships are carried out by the designed neighborhood and the node membership functions which satisfy the minimum conditional threshold. This paper has been inspired by the Join-less approach for mining spatial co-location patterns. We use a spatial database that contains the moving objects and its corresponding spatial location for spatial co-location pattern mining to mine spatially co-located moving objects.

Keywords

Spatial Data Mining, Co-Location, Prim's Algorithm, Moving Objects.

Full Text

     

D. Saravanan ¹, S. Srinivasan ²

Affiliations
1 Sathyabama University, Chennai-119, IN
2 Anna University of Technology, Madurai, IN

Abstract

Video clustering is one of the important tasks in video data mining. We discuss about the different video clustering techniques using hierarchical clustering algorithms like CURE, BIRCH, and CHAMLEON. Set of video frames are clustered using the hierarchical clustering algorithms. As a result of comparison of results all the hierarchical algorithms offer good performance for some set of video files only. That is each clustering algorithm shows best clustering on particular video files only. It fails to forms best clusters of all type of video files. Hence we propose a hierarchical clustering algorithm that offers best clustering performance for all type of video files at any circumstances.

Keywords

Clustering Algorithm, Hierarchical Clustering Algorithm, Video Clustering, Video Data Mining.

Full Text