Difficulty Factors and Preprocessing in Imbalanced Data Sets: An Experimental Study on Artificial Data

Wojciechowski, Szymon; Wilk, Szymon

doi:10.1515/fcds-2017-0007

Cited by 21 publications

(16 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following the same line, Skryjomski & Krawczyk (2017) analyzed the structure of the minority class to transform the SMOTE algorithm into a selective over-sampling method focused on certain types of positive examples. Using two artificial data sets with different dimensions and imbalance ratios, Wojciechowski & Wilk (2017) found out that the critical factor affecting the true-positive rate was the distribution of sample types, while the impact of dimensionality and imbalance ratio was limited. Similarly, Stefanowski (2016) concluded that the performance of the most representative preprocessing approaches depends on the dominating type of minority examples.…”

Section: Distribution-based Data Irregularitiesmentioning

confidence: 99%

Understanding the apparent superiority of over-sampling through an analysis of local information for class-imbalanced data

García

Sánchez

Marqués

et al. 2020

Expert Systems with Applications

View full text Add to dashboard Cite

Data plays a key role in the design of expert and intelligent systems and therefore, data preprocessing appears to be a critical step to produce high-quality data and build accurate machine learning models. Over the past decades, increasing attention has been paid towards the issue of class imbalance and this is now a research hotspot in a variety of fields. Although the resampling methods, either by undersampling the majority class or by over-sampling the minority class, stand among the most powerful techniques to face this problem, their strengths and weaknesses have typically been discussed based only on the class imbalance ratio. However, several questions remain open and need further exploration. For instance, the subtle differences in performance between the over-and under-sampling algorithms are still under-comprehended, and we hypothesize that they could be better explained by analyzing the inner structure of the data sets. Consequently, this paper attempts to investigate and illustrate the effects of the resampling methods on the inner structure of a data set by exploiting local neighborhood information, identifying the sample types in both classes and analyzing their distribution in each resampled set. Experimental results indicate that the resampling methods that pro

show abstract

Section: Distribution-based Data Irregularitiesmentioning

confidence: 99%

Understanding the apparent superiority of over-sampling through an analysis of local information for class-imbalanced data

García

Sánchez

Marqués

et al. 2020

Expert Systems with Applications

View full text Add to dashboard Cite

show abstract

“…[ 4 ] has applied ML to some simple artificial datasets. For rather specific ML questions, artificial data have been used in [ 5 , 6 ].…”

Section: State Of the Artmentioning

confidence: 99%

Generating Artificial Sensor Data for the Comparison of Unsupervised Machine Learning Methods

Zimmering

Niggemann

Hasterok

et al. 2021

Sensors

View full text Add to dashboard Cite

In the field of Cyber-Physical Systems (CPS), there is a large number of machine learning methods, and their intrinsic hyper-parameters are hugely varied. Since no agreed-on datasets for CPS exist, developers of new algorithms are forced to define their own benchmarks. This leads to a large number of algorithms each claiming benefits over other approaches but lacking a fair comparison. To tackle this problem, this paper defines a novel model for a generation process of data, similar to that found in CPS. The model is based on well-understood system theory and allows many datasets with different characteristics in terms of complexity to be generated. The data will pave the way for a comparison of selected machine learning methods in the exemplary field of unsupervised learning. Based on the synthetic CPS data, the data generation process is evaluated by analyzing the performance of the methods of the Self-Organizing Map, One-Class Support Vector Machine and Long Short-Term Memory Neural Net in anomaly detection.

show abstract

“…Data science comprises the preparation, analysis, and processing of both organized and unstructured massive information [16]. Evidence explaining real-world classification difficulties reveals imbalanced distribution in which one of the categories of decisions is underrepresented, often strongly compared to the other class [17]. Larger number of samples from one group would result in a classifier biased to the majority class [18].…”

Section: Data Imbalanced In the Research Scenariomentioning

confidence: 99%

A Modified Adaptive Synthetic SMOTE Approach in Graduation Success Rate Classification

Gameng¹

2019

IJATCSE

View full text Add to dashboard Cite

In the real research situation, the oversampling method in data preprocessing is used to solve the problem in imbalanced data. This imbalance may lessen the capability of classification algorithms to identify instances of interest that lead to misclassification such as false positive generation. These imbalanced datasets come from fields of finance, health, education, among other areas. Academic related data such as graduate success rate on higher education are at times imbalanced. One of the established oversampling methods is the Synthetic Minority Oversampling Technique (SMOTE) with Adaptive Synthetic (Adasyn) SMOTE as one of its many variations. K-Nearest Neighbors (KNN) calculations using Euclidean distance is an embedded in Adasyn. In this study, Manhattan distance is utilized in the KNN calculations. The researchers correspondingly gathered actual data from open admission programs of Davao del Norte State College for the training and testing, which consists of 14 features and 897 records. This modified Adasyn was tested on an imbalanced and primary dataset on graduation success rate using logistic regression and random forest as the classification algorithms. This was evaluated in terms of the performance measurements on overall accuracy, precision, recall, and F1 score. Results showed that the modified Adasyn dominated on each performance metrics over SMOTE and Adasyn. Thus, proving that the modified Adasyn is reliable in decreasing misclassification on the graduate success rate dataset.

show abstract

Difficulty Factors and Preprocessing in Imbalanced Data Sets: An Experimental Study on Artificial Data

Cited by 21 publications

References 28 publications

Understanding the apparent superiority of over-sampling through an analysis of local information for class-imbalanced data

Understanding the apparent superiority of over-sampling through an analysis of local information for class-imbalanced data

Generating Artificial Sensor Data for the Comparison of Unsupervised Machine Learning Methods

A Modified Adaptive Synthetic SMOTE Approach in Graduation Success Rate Classification

Contact Info

Product

Resources

About