A collocation of IRIS flower using neural network clustering tool in MATLAB
V. Poojitha, Madhulika Bhadauria, Shilpi Jain, Anchal Garg · 2016
Background/Objectives: Organizing large amount of data is the biggest challenge in domains of data mining. One way to deal with these kind of issues is by grouping data taking into account the matter of group of files. Clustering is a widely used technique in data mining application for discovering patterns in original data. Many of the algorithms almost all customized clustering algorithms are restricted in taking care and managing datasets that consists of unconditional characteristics and attributes. However, datasets with unconditional types of attributes are common in real life data mining problem. Clustering is a fundamental data analysis method. It is a way to form natural groupings in the given set of data. K-means is a simple unsupervised clustering techniques. Neural network clustering tool is best for obtaining optimal clustering of large data set as it uses unsupervised competitive technique and clusters by liner dicrimant. It is a form of unsupervised learning where generally we do not have examples demonstrating how the data should be grouped together [7]. Methodology: Existing iris flower dataset is preloaded in MATLAB and is used for clustering into three different species. The dataset is clustered using the k-means algorithm and neural network clustering tool in MATLAB. Neural network clustering tool is mainly used for clustering large data set without any supervision. It is also used for pattern recognition, feature extraction, vector quantization, image segmentation, function approximation, and data mining. Results/Findings: The results include the clustered iris dataset into three species without any supervision. The neighboring cluster's/neuron's distances and weights have also been plotted using self organizing map for better understanding of unsupervised clustering. As an unsupervised classification technique, clustering - 2 - identifies some inherent structures present in a set of objects based on a similarity measure. Clustering techniques are formed on the basis of statistical model identification or competitive learning. The research paper shows detailed outline of competitive learning formed on clustering methods like k-means and obtaining optimal clustering of iris dataset using neural network clustering tool.