Enhanced Data Cleaning and Outlier Detection Models with Hybrid Metaheuristic Approaches

Abstract

Applications run based on clean data analysis reports. Data are generated through newlinedifferent sources such as social media, e-commerce, blogs, banking, healthcare, transactions, newlineapps, websites, opinion platforms, etc. So, there is a chance of dirty data increase. It is newlineprocessed for effective utilization in different industries, including healthcare. Every newlinebusiness enterprises require noise-free and clean data. Dirty data includes incorrect data, newlinedummy data, duplicate data, inappropriate data, violation of data, etc. Due to different newlinecauses, including abbreviation misuse, incorrect spelling, outdated data, and dirty data newlinegenerated. To avoid the wrong conclusions, the data cleaning process becomes vital. The newlineeffort is to introduce a novel data-cleaning technique to remove dirty data effectively. This newlineprocess involves (i) Dirty data detection and (ii) Dirty data cleaning. newlineThe dirty data detection process has been assigned the following: data normalization, newlinehashing, clustering, and finding the suspected data. In the clustering process, the optimal newlinecentroid selection is promising and is carried out by employing the optimization concept. newlineAfter the finishing of dirty data prediction, the subsequent process of dirty data cleaning newlinebegins to activate. The cleaning process also assigns some processes: the leveling process, newlineHuffman coding, and cleaning the suspected data. The cleaning of suspected data is newlineperformed based on the optimization concept. newlineHence, to solve all optimization problems, a new hybridized algorithm is developed newlinecalled the Firefly Update enabled Rider Optimization Algorithm (FU-ROA), which is the newlinehybridization of the Rider Optimization Algorithm (ROA) and Firefly (FF) algorithms. In newlinethe end, the analysis of the performance of the implanted data cleaning method is compared newlineover the other traditional methods like Particle Swarm Optimization (PSO), FF, Grey Wolf newlineOptimizer (GWO), and ROA in terms of their positive and negative measures.