In high-dimensional data analysis, Feature Selection (FS) is one of the most fundamental issues in machine learning and requires the attention of researchers. These datasets are characterized by huge space due to a high number of features, out of which only a few are significant for analysis. Thus, significant feature extraction is crucial. There are various techniques available for feature selection; among them, the filter techniques are significant in this community, as they can be used with any type of learning algorithm and drastically lower the running time of optimization algorithms and improve the performance of the model. Furthermore, the application of a filter approach depends on the characteristics of the dataset as well as on the machine learning model. Thus, to avoid these issues in this research, a combination of feature reduction (CFR) is considered designing a pipeline of filter approaches for high-dimensional microarray data classification. Considering four filter approaches, sixteen combinations of pipelines are generated. The feature subset is reduced in different levels, and ultimately, the significant feature set is evaluated. The pipelined filter techniques are Correlation-Based Feature Selection (CBFS), Chi-Square Test (CST), Information Gain (InG), and Relief Feature Selection (RFS), and the classification techniques are Decision Tree (DT), Logistic Regression (LR), Random Forest (RF), and k-Nearest Neighbor (k-NN). The performance of CFR depends highly on the datasets as well as on the classifiers. Thereafter, the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method is used for ranking all reduction combinations and evaluating the superior filter combination among all.

The classification of frequent item set is a vital problem in data mining. In recent area, where it deals with more outcome within limited time, there is a need of certain algorithm supported by faster computation. The aim of this research work is to generate a frequent item set so as to achieve a comparatively reduced time and effort. Frequent item set is a set of items that is often bought with another item, for example bread and butter. Here, two association rules mining algorithm, that is, Apriori and Eclat are used to predict the probability of subordinate items being bought if a prime item is obtained. Apriori algorithm works by picking the frequent individual items in a database and extending them to larger and larger sets whereas Eclat algorithm uses depth first search for discovering frequent item sets. In this paper, a dataset of transactions in a retail company is taken and corresponding support, confidence and lift is estimated. Here the relationship between these variables is explored. It is observed in Eclat algorithm that even after varying the minimum support the maximum and minimum support remains constant and using the Apriori algorithm that increasing the minimum lift value results in a gradual decrease in maximum support and small increase in minimum confidence values.

Optical Character Recognition (OCR) is a document image analysis method where scanned digital image that contains either machine printed or handwritten scripts are input into a system to translate it into an editable machine readable digital text format. Hence OCR has been a highly challenging topic of interest for researchers all around the globe and also research paper involving OCR that works under all possible conditions and gives highly accurate results. It is seen that efficient algorithms have increased the speed and accuracy of character recognition. A substantial amount of work has been done on foreign languages such as English, Chinese etc. but very few paper are there for Indian languages baring a few for Hindi and Bengali. Hence our research work was directed towards development of a novel algorithm for Offline Typewritten Odia Character Recognition using Template Matching with Unicode Mapping.

Data mining is the process of obtaining knowledge and information from massive amounts of data. Data mining is mostly used for data analysis. In Data Mining various techniques are used that are association mining, regression, prediction, classification, clustering, etc. Classification is described as the process of identifying a collection of models (or functions) that explain and differentiate data classes and ideas, to use the model for detecting the classes of unknown objects or patterns, whose class designations aren't clear. Classification is a supervised learning problem. That means in machine learning, Classification is the problem of identifying data patterns from the group of patterns according to their characteristics and define which patterns are from which class. Pattern classification can be done by using various classifiers. A classifier is a program that inputs the feature vector of a pattern or data point and assigns it to one of a set of designated classes. The classifiers such as Artificial Neural Network (ANN), k-Nearest Neighbor (k-NN) classifier, Support Vector Machine (SVM), etc. are used for pattern classification purposes. Focusing on the classification technique of data mining, in this research work the accuracy of k-NN using three datasets from the UCI machine learning library is presented. The main goal of this paper is to provide a review to find out the accuracy of the k-NN classification technique using different datasets in data mining. The k-NN classifier is a simple but efficient approach used for classification in research.

Phrase mining is the way toward deriving aspects of expressions from the collection of texts. Several uses of phrase mining include information retrieval/extraction, taxonomy construction, and topic modeling. The existing strategy requires a prepared linguistic analyzer and has an unacceptable execution for new areas since it requires human specialists for labeling the phrase. The phrase generated in those systems for a given input text contains only a single word that may be often unambiguous to the user. The aim is to automate the phrase mining process and enhance its performance. The proposed method is a framework that requires minimal human labeling effort and only shallow linguistic analysis. A POS_tagger is used to extract the important words (nouns and noun phrases) from a text after which text ranking is applied. Then cosine similarity is used to identify the quality phrase from the text. Phrase quality can be estimated at two levels, once after POS_guided segmentation and then re-estimate the score at the end. Compared to the existing method, the proposed method has showna significant improvement in effectiveness and efficiency across different domains. This technique can be reached out to support any language up to a normal learning base (e.g. Wikipedia) of comparing vocabulary is accessible.