Search

Your search keyword '"Gao, Songfeng"' showing total 13 results
Start Over Author "Gao, Songfeng" Publication Year Range Last 3 years
13 results on '"Gao, Songfeng"'

1. A comparative study on machine learning models combining with outlier detection and balanced sampling methods for credit scoring

Author

Qian, Hongyi, Zhang, Shen, Wang, Baohui, Peng, Lei, Gao, Songfeng, and Song, You

Subjects
Computer Science - Machine Learning
Abstract

Peer-to-peer (P2P) lending platforms have grown rapidly over the past decade as the network infrastructure has improved and the demand for personal lending has grown. Such platforms allow users to create peer-to-peer lending relationships without the help of traditional financial institutions. Assessing the borrowers' credit is crucial to reduce the default rate and benign development of P2P platforms. Building a personal credit scoring machine learning model can effectively predict whether users will repay loans on the P2P platform. And the handling of data outliers and sample imbalance problems can affect the final effect of machine learning models. There have been some studies on balanced sampling methods, but the effect of outlier detection methods and their combination with balanced sampling methods on the effectiveness of machine learning models has not been fully studied. In this paper, the influence of using different outlier detection methods and balanced sampling methods on commonly used machine learning models is investigated. Experiments on 44,487 Lending Club samples show that proper outlier detection can improve the effectiveness of the machine learning model, and the balanced sampling method only has a good effect on a few machine learning models, such as MLP.

Published
2021

2. Managing dataset shift by adversarial validation for credit scoring

Author

Qian, Hongyi, Wang, Baohui, Ma, Ping, Peng, Lei, Gao, Songfeng, and Song, You

Subjects
Computer Science - Machine Learning
Abstract

Dataset shift is common in credit scoring scenarios, and the inconsistency between the distribution of training data and the data that actually needs to be predicted is likely to cause poor model performance. However, most of the current studies do not take this into account, and they directly mix data from different time periods when training the models. This brings about two problems. Firstly, there is a risk of data leakage, i.e., using future data to predict the past. This can result in inflated results in offline validation, but unsatisfactory results in practical applications. Secondly, the macroeconomic environment and risk control strategies are likely to be different in different time periods, and the behavior patterns of borrowers may also change. The model trained with past data may not be applicable to the recent stage. Therefore, we propose a method based on adversarial validation to alleviate the dataset shift problem in credit scoring scenarios. In this method, partial training set samples with the closest distribution to the predicted data are selected for cross-validation by adversarial validation to ensure the generalization performance of the trained model on the predicted samples. In addition, through a simple splicing method, samples in the training data that are inconsistent with the test data distribution are also involved in the training process of cross-validation, which makes full use of all the data and further improves the model performance. To verify the effectiveness of the proposed method, comparative experiments with several other data split methods are conducted with the data provided by Lending Club. The experimental results demonstrate the importance of dataset shift in the field of credit scoring and the superiority of the proposed method.

Published
2021

5. Managing Dataset Shift by Adversarial Validation for Credit Scoring

Author

Qian, Hongyi, Wang, Baohui, Ma, Ping, Peng, Lei, Gao, Songfeng, Song, You, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Khanna, Sankalp, editor, Cao, Jian, editor, Bai, Quan, editor, and Xu, Guandong, editor

Published
2022
Full Text
View/download PDF

11. Reconstruction of Annual Glacier Mass Balance from Remote Sensing-Derived Average Glacier-Wide Albedo.

Author

Zhang, Zhimin, Jiang, Liming, Sun, Yafei, Sirguey, Pascal, Dumont, Marie, Liu, Lin, Gao, Ning, and Gao, Songfeng

Subjects
MASS budget (Geophysics), ALBEDO, GLACIERS, STANDARD deviations, DIGITAL elevation models
Abstract

Annual mass balance is an important reflection of glacier status that is also very sensitive to climate fluctuations. However, there is no effective and universal albedo-based method for the reconstruction of annual mass balance due to the scarcity of field observations. Here, we present an improved albedo–mass balance (IAMB) method to estimate annual glacier surface mass balance series using remote sensing techniques. The averaged glacier-wide albedo derived with the MODImLab algorithm during the summer season provides an effective proxy of the annual mass change. Defined as the variation in the albedo as a function of elevation change, the altitude–albedo gradient ( ∂ z / ∂ α ) can be obtained from a glacier digital elevation model (DEM) and optical images. The Chhota Shigri glacier situated in the western Himalayas was selected to test and assess the accuracy of this method over the period from 2003 to 2014. Reconstructed annual mass budgets correlated well with those from the observed records, with an average difference and root mean square error (RMSE) of −0.75 mm w.e. a−1 and 274.91 mm w.e. a−1, respectively, indicating that the IAMB method holds promise for glacier mass change monitoring. This study provides a new technique for annual mass balance estimation that can be applied to glaciers with no or few mass balance observations. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

13. Soil organic carbon fractions and its association with water‐stable aggregates under different fertilization management practices.

Author

Zhao, Zhanhui, Gao, Songfeng, Lu, Chunyang, Li, Xiaoyu, and Wang, Tingyun

Subjects
CARBON in soils, FERTILIZER application, FERTILIZERS, SOIL structure, CARBON sequestration, MANURES
Abstract

Agricultural soil organic carbon (SOC) plays a critical role in soil ecosystems, and SOC sequestration is a key ecosystem service. This study investigated the SOC fractions, including mineral‐incorporated organic carbon (MSC), intra‐particulate organic carbon (iPOC) and microbial biomass carbon (MBC), in bulk soil and water‐stable aggregates, as influenced by soil fertilization managements, using three treatments: no fertilizer (CK), chemical fertilizer (FR) and organic manure (OM). Fertilizer application significantly increased SOC content and its fractions (MSC, iPOC and MBC) compared to CK, and the effect of organic manure amendment had more impact on the SOC accumulation than chemical fertilizer. Macroaggregate (>250 μm)‐associated SOC, MSC and iPOC (including fine and coarse iPOC) increased by 22%–283%, 18%–84% and 45%–232%, respectively, in FR and OM treatments compared with CK. In addition, the iPOC:MSC ratio values (%) within all size aggregates were all increased in the FR (about 84%–183%) and OM (about 103%–230%) treatments compared to CK (about 65%–146%). Meanwhile, MBC:SOC ratio values were the highest (about 2.16%–3.07%) in small macroaggregates (250–2000 μm), but decreased sharply (about 2%–2.35%) in large macroaggregates (>2000 μm) in different treatments. Our findings suggested that macroaggregates were the main sites of carbon sequestration and that carbon became more unstable with the increase of SOC in aggregates. Furthermore, the microbial biomass showed obvious differentiation characteristics in macroaggregates (>250 μm). Specifically, small macroaggregates (250–2000 μm) contained more microbial biomass, and large macroaggregates (>2000 μm) were inclined to inhibit microbial activity. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results