A New Heuristic Approach for Treating Missing Value: ABCimp

Abstract

Missing values in datasets present an important problem for traditional and modern statistical methods. Many statistical methods have been developed to analyze the complete datasets. However, most of the real world datasets contain missing values. Therefore, in recent years, many methods have been developed to overcome the missing value problem. Heuristic methods have become popular in this field due to their superior performance in many other optimization problems. This paper introduces an Artificial Bee Colony algorithm based new approach for missing value imputation in the four real-world discrete datasets. At the proposed Artificial Bee Colony Imputation (ABCimp) method, Bayesian Optimization is integrated into the Artificial Bee Colony algorithm. The performance of the proposed technique is compared with other well-known six methods, which are Mean, Median, k Nearest Neighbor (k-NN), Multivariate Equation by Chained Equation (MICE), Singular Value Decomposition (SVD), and MissForest (MF). The classification error and root mean square error are used as the evaluation criteria of the imputation methods performance and the Naive Bayes algorithm is used as the classifier. The empirical results show that state-of-the-art ABCimp performs better than the other most popular imputation methods at the variable missing rates ranging from 3 % to 15 %.

Missing Value Imputation Designs and Methods of Nature-Inspired Metaheuristic Techniques: A Systematic Review

“…Multi objective approaches Hybrid approaches GA [16]- [18] GP [19] PSO [20] MOGA-II [21]- [22] MOPSO [23] Bayesian ACO+ Bayesian [24] ABC+ Bayesian [25] Max-min ACO +bayesian [26]- [27] Bayesian+ tensor+chaotic PSO [28] Probabilistic GA+KNN [29] GMSA+MPSO+ WKNN [30] PSO+ covariance matrix [32] IDW+TR+ PSO [33] Clustering ACO+ clustering [34] FCM+GA [35][36] FCM+ SVR+GA [37] GA+SOM [38] FCM+PSO [39]- [42] GFM+PSO [43] PSO-ECM+ AAELM [44] ELM+PSO+ FCM [45] PSO+K-means+ ontology [46] SOM+FOA +LSSVM [47] DE+ clustering [48] GA+RF [49] GP+wrapper [55] [56] Neural network GSO+MLP [57] GA+MLP, SA+MLP, PSO+MLP, RF+MLP [58] SC-FITNET [59] SC-FDO+ MLP [60] DL-CS [61] DL-BAT [62] DL-GSA [63] PSO+LSVM [54] PSO+levy flight+SVM [53] MAIS+GA [50] GA+ARO [51] GP+tree vector [52] KNN+LAHC AWOA [31] The proposed approach enhanced imputation for missing multivariate data…”

“…However, nearly 14.6% of the studies did not reveal their missing rates for the experimentation. The works in [24]- [25], for example, used the Framingham heart dataset with real missing values, but the authors did not disclose the proportion of missing values in the dataset. Nevertheless, the missing rates greater than 50% category received the least attention, accounting for 14.6% (7/48) of the studies.…”

“…Another standard metric is error for summarizing the performance of imputation and classification models. For example, researchers adopted error metric to measure the classification errors of the proposed imputation models in the missing iris dataset [22], [35], [39], poker hand dataset [24], [26], website phishing dataset [25] and health datasets [31]. On the other hand, RA is an indicator of how many estimations fail within a standard range [36], [37], [45], [61], [62], while specificity (also true negative rate) refers to the proportion of sample without the condition but obtained a negative result [18], [23], [28], [43].…”

“…Most studies reported that at lower missingness, the performances of MVI are relatively better. Imputation errors increased when missing rates increased, for examples in [22], [24], [25], [28], [33], [34], [45], [46]. In addition, the percentages of missingness greatly influenced the work in [26], [35].…”

Estimating The Aeration Performance Of Venturi-Conduit By Artificial Bee Colony Programming

“…ABC Algorithm simulates e food source search strategy of a honey bee swarm [12]. Due to its fast and strong convergence ability, besides many optimization problems in computer science such as feature selection, intrusion detection, clustering, missing value imputation, it is also applied to optimize many engineering problems [13][14][15][16]. Disributed generation allocation problem, composite design optimization problem, photovoltaic parameter estimation problem, flow shop scheduling problem, estimation of the fundamental period of vibration are just some of the application areas.…”