Performance evaluation of outlier detection techniques in production timeseries: A systematic review and meta-analysis

Alimohammadi, Hamzeh; Chen, Shengnan Nancy

doi:10.1016/j.eswa.2021.116371

Cited by 54 publications

(17 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section we discuss and analyze the results of running the five machine learning algorithms, mainly k-NN (k-nearestneighbor) -a nearest-neighbor-based unsupervised algorithm focused on detection of global anomalies (global relative to the dataset) with low computational impact [47], LOF (Local Outlier Factor) -a nearest-neighbor-based algorithm able to detect local anomalies alongside global ones [40], [48], LOCI (Local Correlation Integral) -a nearest-neighbor-based local algorithm with increased precision over k-NN but also with increased computational complexity [43], [45], CBLOF (cluster-based local outlier factor) -a clustering-based global algorithm [49] and HBOS (histogram-based outlier score), a very fast statistical algorithm almost an order of magnitude faster than k-NN [50].…”

Section: Discussionmentioning

confidence: 99%

A Comparative Study of Unsupervised Anomaly Detection Algorithms used in a Small and Medium-Sized Enterprise

Petrariu¹,

Moscaliuc²,

Turcu³

et al. 2022

IJACSA

View full text Add to dashboard Cite

Anomaly detection finds application in several industries and domains. The anomaly detection market is growing driven by the increasing development and dynamic adoption of emerging technologies. Depending on the type of supervision, there are three main types of anomaly detection techniques: unsupervised, semi-supervised, and supervised. Given the wide variety of available anomaly detection algorithms, how can one choose which approach is most appropriate for a particular application? The purpose of this evaluation is to compare the performance of five unsupervised anomaly detection algorithms applied to a specific dataset from a small and medium-sized software enterprise, presented in this paper. To reduce the cost and complexity of a system developed to solve the problem of anomaly detection, a solution is to use machine learning (ML) algorithms that are available in one of the open-source libraries, such as the scikit-learn library or the PyOD library. These algorithms can be easily and quickly integrated into a low-cost software application developed to meet the needs of a small and medium-sized enterprise (SME). In our experiments, we considered some unsupervised algorithms available in PyOD library. The obtained results are presented, alongside with the limitations of the research.

show abstract

Section: Discussionmentioning

confidence: 99%

A Comparative Study of Unsupervised Anomaly Detection Algorithms used in a Small and Medium-Sized Enterprise

Petrariu¹,

Moscaliuc²,

Turcu³

et al. 2022

IJACSA

View full text Add to dashboard Cite

show abstract

“…The presence of outliers might result in the uncertainties in forecast as well as inaccurate predictions [61]. Therefore, detecting and removing them are of great importance to increase the data quality and the performance of the predictive models.…”

Section: Proposed Scheme Descriptionmentioning

confidence: 99%

A combination of novel hybrid deep learning model and quantile regression for short‐term deterministic and probabilistic PV maximum power forecasting

Tuyen

Thanh

Huu

et al. 2022

IET Renewable Power Gen

View full text Add to dashboard Cite

Recently, the increasing penetration of renewable energy resources in power system, especially photovoltaic (PV) systems, has caused severe technical issues due to its randomness and dependency on primary sources. Therefore, precise output power forecast is vital for both system operators and PV system owners to improve grid stability and generation quality, respectively. This article proposes a novel hybrid deep learning model named EDSACL comprising four components, namely, convolutional neural network (CNN), long short‐term memory (LSTM), self‐attention mechanism (SAm), and residual learning (RL) strategy for short‐term PV maximum power prediction with forecast horizon of 1, 2, and 4 h. This proposed model initially extracts spatial‐temporal features of input data based on CNN and LSTM before using SAm to achieve hidden features of LSTM layers. RL strategy is employed to maintain the information flow entering the network. Besides, interval prediction is also implemented based on a combination of the EDSACL model and quantile regression with different probabilities. The forecast accuracy of the proposed model is validated based on two real‐world data sets. The predicted results show the superiority of the proposed model over numerous predictive models; in particular, mean absolute percentage error values of the proposed hybrid model are under 6% in all case studies.

show abstract

“…Several authors, such as Markou and Singh [11], Goldstein et al [12], Patcha et al [13], and Alimohammadi et al [14], have published surveys and reviews about outlier detection methods, some of which provide an analytical comparison of the features of existing outlier detection methods. In this context, LOF and DBSCAN are among the most popular solutions for outlier detection.…”

Section: Related Workmentioning

confidence: 99%

An Analysis of ML-Based Outlier Detection from Mobile Phone Trajectories

Pereira

Sofia

2022

Future Internet

View full text Add to dashboard Cite

This paper provides an analysis of two machine learning algorithms, density-based spatial clustering of applications with noise (DBSCAN) and the local outlier factor (LOF), applied in the detection of outliers in the context of a continuous framework for the detection of points of interest (PoI). This framework has as input mobile trajectories of users that are continuously fed to the framework in close to real time. Such frameworks are today still in their infancy and highly required in large-scale sensing deployments, e.g., Smart City planning deployments, where individual anonymous trajectories of mobile users can be useful to better develop urban planning. The paper’s contributions are twofold. Firstly, the paper provides the functional design for the overall PoI detection framework. Secondly, the paper analyses the performance of DBSCAN and LOF for outlier detection considering two different datasets, a dense and large dataset with over 170 mobile phone-based trajectories and a smaller and sparser dataset, involving 3 users and 36 trajectories. Results achieved show that LOF exhibits the best performance across the different datasets, thus showing better suitability for outlier detection in the context of frameworks that perform PoI detection in close to real time.

show abstract

Performance evaluation of outlier detection techniques in production timeseries: A systematic review and meta-analysis

Cited by 54 publications

References 19 publications

A Comparative Study of Unsupervised Anomaly Detection Algorithms used in a Small and Medium-Sized Enterprise

A Comparative Study of Unsupervised Anomaly Detection Algorithms used in a Small and Medium-Sized Enterprise

A combination of novel hybrid deep learning model and quantile regression for short‐term deterministic and probabilistic PV maximum power forecasting

An Analysis of ML-Based Outlier Detection from Mobile Phone Trajectories

Contact Info

Product

Resources

About