Abstract SARS‐Coronavirus was first detected in December 2019, later named COVID‐19, and declared a pandemic by the World Health Organization (WHO). As prediction models assist policymakers in making decisions based on expected outcomes. Existing models were only used to anticipate a smaller range of data resulting in irrelevant predictions. Our research focuses on predicting COVID‐19 confirmed, recovered, and deceased Indian cases for 20 days ahead. Tuning of hyperparameters is performed with a grid search cross‐validation approach. The dataset is collected from the Kaggle. Our forecast indicates that the count of confirmed and deceased cases is higher whereas, recovered cases prediction shows a decreasing trend. The R2 Score achieved is 0.5112 and root‐mean‐square error (RMSE) is 1251 using optimized SARIMAX. Finally, Monte Carlo simulation has also been performed to justify the prediction accuracy as compared to other models such as linear, polynomial, prophet, and SARIMAX without grid search cross validation.

Summary Optimal feature selection of a high‐dimensional micro‐array datasets has gained a significant importance in medical applications for early detection and prevention of disease. Traditional Optimal feature selection percolates through a population‐based meta‐heuristic optimization technique, a Machine Learning classifier and traditional wrapper method for transforming the original feature set into a better feature set. These techniques require a number of iterations for the convergence of random solutions to the global optimum with high‐dimensionality issues such as over‐fitting, memory constraints, computational costs, and low accuracy. In this article, an efficient equilibrium optimization technique is proposed for an optimized feature selection that increases the diversity of the population in the search space through Random Opposition based learning and classify the best features using a 10‐fold cross‐validation‐based wrapper method. The proposed method is tested with six standard micro‐array datasets and compared with the conventional algorithms such as Marine Predators Algorithm, Harris Hawks Optimization, Whale Optimization Algorithm, and conventional Equilibrium Optimization. From the statistical results using the standard metrics, it is interpreted that the proposed method converges to the global minimum in a few iterations through optimized feature selection, fitness value and higher classification accuracy. This proves its efficacy in exploring and finding a better solution as compared to the counterpart algorithms. In addition to complexity analysis, these results indicate a global optimum solution, an effective representation of least amount of data‐high dimensionality reduction and an avoidance of over‐fitting problems. The source code is available at https://github.com/balasv/ROBL‐EOA/blob/main/ROBL_EOA.ipynb

Summary Obtaining precise information from a high‐dimensional dataset is one of the most difficult tasks as datasets contain more features and fewer samples. The high‐dimensionality of the dataset reduces predictive capability and increases the computational complexity of the analytical model. The widespread employment of meta‐heuristic methods to handle the challenge of high‐dimensional datasets has been exceptional in recent years. The marine predators algorithm (MPA) is a recently developed meta‐heuristic algorithm based on the “survival‐of‐the‐fittest” notion. This research critique overcomes the drawbacks of the existing MPA and proposes a feature selection model using random opposition‐based learning (ROBL). The searching for the optimum solution in a single direction of the traditional MPA reduces its performance. The incorporation of ROBL in the MPA enhances its ability to reconnoiter bigger search space. The proposed algorithm generates a new population based on the initial and random opposite population. The performance of ROBL‐MPA is inspected on six high‐dimensional microarray datasets. The results of the proposed ROBL‐MPA are compared to traditional MPA and opposition based MPA (OBL‐MPA). The proposed ROBL‐MPA outperforms traditional MPA based on several benchmark performance analysis tests.

Abstract Retrieving the relevant information from the high‐dimensional dataset enhances the classification accuracy of a predictive model. This research critique has devised an improved marine predator algorithm based on opposition learning for stable feature selection to overcome the problem of high‐dimensionality. Marine predator algorithm is a population‐based meta‐heuristics optimization algorithm that works on the ‘survival‐of‐the‐fittest’ theory. Classical marine predator algorithm explores the search space merely in one direction, affecting its converging capacity while being responsible for stagnation at local minima. The proposed opposition‐based learning nuances enhance the exploration capacity of marine predator algorithm and productively converges the model to global optima. The proposed OBL‐based marine predator algorithm selects stable, substantial elements from six different high‐dimensional microarray datasets. The performance of the proposed method is investigated using five predominantly used classifiers. From the result, it is understood that the proposed approach outperforms other conventional feature selection techniques in terms of converging capability, classification accuracy, and stable feature selection.

Abstract The African vulture optimization algorithm (AVOA) is a recently developed metaheuristic algorithm that imitates the eating and movement patterns of authentic African vultures. AVOA is developed to address the continuous optimization problem. However, AVOA is unable to solve the discrete search space, this inspires us to develop the binary AVOA for feature selection problems in classification tasks. The suggested BAVOA incorporates an eight‐transfer function (S‐shaped and V‐shaped) for transforming a continuous variable to a binary one. Using 14 benchmark data sets, the proposed technique is compared against 15 conventional binary metaheuristics algorithms in terms of classification accuracy, fitness function, number of selected features and converging ability. Furthermore, the results are statistically analysed using Wilcoxon test. The comparative findings of S‐Shaped and V‐shaped transfer functions indicate the superior performance of BAVOA methods, particularly S2‐BAVOA, in contrast to other transfer function. Based on results, it turns out that the suggested technique converges to the global minimum in several iterations based on the selection of optimal characteristics, fitness values and higher classification accuracy as compared to the classical binary metaheuristic algorithms.

For optimal results, retrieving a relevant feature from a microarray dataset has become a hot topic for researchers involved in the study of feature selection (FS) techniques. The aim of this review is to provide a thorough description of various, recent FS techniques. This review also focuses on the techniques proposed for microarray datasets to work on multiclass classification problems and on different ways to enhance the performance of learning algorithms. We attempt to understand and resolve the imbalance problem of datasets to substantiate the work of researchers working on microarray datasets. An analysis of the literature paves the way for comprehending and highlighting the multitude of challenges and issues in finding the optimal feature subset using various FS techniques. A case study is provided to demonstrate the process of implementation, in which three microarray cancer datasets are used to evaluate the classification accuracy and convergence ability of several wrappers and hybrid algorithms to identify the optimal feature subset.

Abstract A microarray dataset contains thousands of DNA spots covering almost every gene in the genome. Microarray-based gene expression helps with the diagnosis, prognosis and treatment of cancer. The nature of diseases frequently changes, which in turn generates a considerable volume of data. The main drawback of microarray data is the curse of dimensionality. It hinders useful information and leads to computational instability. The main objective of feature selection is to extract and remove insignificant and irrelevant features to determine the informative genes that cause cancer. Random forest is a well-suited classification algorithm for microarray data. To enhance the importance of the variables, we proposed out-of-bag (OOB) cases in every tree of the forest to count the number of votes for the exact class. The incorporation of random permutation in the variables of these OOB cases enables us to select the crucial features from high-dimensional microarray data. In this study, we analyze the effects of various random forest parameters on the selection procedure. ‘Variable drop fraction’ regulates the forest construction. The higher variable drop fraction value efficiently decreases the dimensionality of the microarray data. Forest built with 800 trees chooses fewer important features under any variable drop fraction value that reduces microarray data dimensionality.

Abstract With the advent of the internet, the recommender system escorts the users in a customized way to nominate items from a massive set of possible alternatives. The emergence of overspecification in recommender system has emphasized negative effects on the context of prediction. The drift of user interest over time is one of the challenging affairs in present personalized recommender system. In this paper, we present a neural network model to improve the recommendation performance along with usage of fuzzy-based clustering to decide membership value of users and matching imputation to cutback sparsity to some extent. We evaluate our model on the MovieLens dataset and show that our model not only elevates accuracy, but also considers the order in which recommendation should be given. We compare the proposed model with a number of state-of-the-art personalization methods and show the dominance of our model using accuracy metrics such as root-mean-square error and mean absolute error.