Restructuring performance prediction with a rebalanced and clustered support vector machine

Li, Hui; Hong, Lu‐Yao; Mo, Yuchang; Zhu, Bang‐Zhu; Chang, Pei‐Chann

doi:10.1002/for.2512

Cited by 8 publications

(3 citation statements)

References 70 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[42] proposed a KNN-based oversampling strategy to generate minority class samples for Chinese tourism business failure warnings. [43] combined clustering-based under-sampling strategy and a boosting ensemble framework to business failure prediction, and provided reference value to fraud detection, credit scoring related domain. [44] incorporated SMOTE algorithm into a Geometric mean-aware boosting framework to alleviate the class imbalance problem in the bankruptcy prediction task.…”

Section: B Imbalanced Business Failure Predictionmentioning

confidence: 99%

Business Failure Prediction Based on a Cost-Sensitive Extreme Gradient Boosting Machine

Zou

Gao

2022

IEEE Access

View full text Add to dashboard Cite

Business failure prediction is very important for the sustainable development of enterprises. Machine learning algorithms, especially ensemble algorithms, have shown great economic benefits in enterprise financial early warning. However, the highly imbalanced class distribution of financial risk data and the inexplainable of most machine learning-based early distress warning models limit their commercial application. To address the above limitations, we enhance the business failure prediction performance by tree-ensemble in a boosting manner. Moreover, to solve the class imbalanced issue in business failure datasets, a weighted objective function, weighted cross-entropy, is embedded into the boosted tree framework, making the weighted XGBoost a cost-sensitive business failure prediction model. Besides, to tackle the second issue, we explore the intrinsic interpretability of the proposed method by visualizing the feature importance and incorporating a partial dependence plot technique to locally interpret the individual business failure event. Experimental results on business failure datasets with different predictive horizons collected from China Security Market Accounting Research Database (CSMARD) show the proposed weighted XGBoost is a good solution to reduce the error on recognizing firms in business failure. Furthermore, the visualized feature importance score and partial dependence plot result both demonstrate that the cost-sensitive tree-based ensemble can be a good tool to guide the investors in making rational as well as provide interpretable business prediction results as a reference for the policymaking of the regulators.

show abstract

Section: B Imbalanced Business Failure Predictionmentioning

confidence: 99%

Business Failure Prediction Based on a Cost-Sensitive Extreme Gradient Boosting Machine

Zou

Gao

2022

IEEE Access

View full text Add to dashboard Cite

show abstract

“…In addition to performing linear classification, SVM can effectively perform nonlinear classification using so-called kernel techniques, implicitly mapping its inputs to highdimensional feature spaces. There are some studies using the SVM to predict the financial data [45,46].…”

Section: ) Support Vector Machine (Svm)mentioning

confidence: 99%

Stock Trend Prediction Using Candlestick Charting and Ensemble Machine Learning Techniques With a Novelty Feature Engineering Scheme

et al. 2021

View full text Add to dashboard Cite

Stock market forecasting is a knotty challenging task due to the highly noisy, nonparametric, complex and chaotic nature of the stock price time series. With a simple eight-trigram feature engineering scheme of the inter-day candlestick patterns, we construct a novel ensemble machine learning framework for daily stock pattern prediction, combining traditional candlestick charting with the latest artificial intelligence methods. Several machine learning techniques, including deep learning methods, are applied to stock data to predict the direction of the closing price. This framework can give a suitable machine learning prediction method for each pattern based on the trained results. The investment strategy is constructed according to the ensemble machine learning techniques. Empirical results for the period of 2000 to 2017 of China's stock market confirm that our feature engineering has effective predictive power, with a prediction accuracy of more than 60% for some trend patterns. Various measures such as big data, feature standardization, and elimination of abnormal data can effectively solve the problem of data noise. An investment strategy based on our forecasting framework excels in both individual stock and portfolio performance theoretically. Transaction costs have a significant impact on investment. Additional technical indicators can improve the forecast accuracy to varying degrees. Technical indicators, especially the momentum indicators, can improve forecasting accuracy in most cases.

show abstract

“…Bike sharing system has a large number of stations and complex attributes, which makes it difficult to calibrate their attributes one by one. However, due to the self-fluidity of shared bicycle [4,5], there is a strong correlation between some stations.Therefore, if the association rules are used to collect stations with strong correlation and adopt K-medoids algorithm [6,7] and constraint adjustment for scheduling, it can effectively avoid large random errors.…”

Section: Introductionmentioning

confidence: 99%

Research on Scheduling Area Partition Method Based on Multiple Algorithms

Li¹,

Wang²

2019

International Conferences on Software Engineering and Knowledge Engineering

View full text Add to dashboard Cite

Shared bicycle, as a green means of transportation, is very popular among people and it is an important way for many people to travel daily. In recent years, with the increasing scale and frequency of bike sharing system, the unbalanced use of shared bicycle has a great impact on the users' experience, which is one of the main problems faced by current system operators. of Division of the traffic area can not only provide a new idea for solving the problem of unbalanced bike usage, but also provide a theoretical and practical basis for the planning, layout, construction, operation and scientific dispatching of shared bicycle system. However, there are few clear methods to study the partition method of shared dispatching area. To solve this problem, based on historical bicycle data, traffic station data, we analyze the rules of shared bicycle space-time characteristics and propose a method of dividing shared bicycle dispatching areas by combining K-medoids clustering, association rules and total demand constraint adjustment. We evaluate our approach on the New York City (NYC Citi Bike) bicycle sharing system and show the advantages of our approach for Large-scale station-level dispatching area optimization (beyond baseline approaches).

show abstract

Restructuring performance prediction with a rebalanced and clustered support vector machine

Cited by 8 publications

References 70 publications

Business Failure Prediction Based on a Cost-Sensitive Extreme Gradient Boosting Machine

Business Failure Prediction Based on a Cost-Sensitive Extreme Gradient Boosting Machine

Stock Trend Prediction Using Candlestick Charting and Ensemble Machine Learning Techniques With a Novelty Feature Engineering Scheme

Research on Scheduling Area Partition Method Based on Multiple Algorithms

Contact Info

Product

Resources

About