arXiv subjects

Nanbo Peng

Publications and source records attributed to Nanbo Peng.

6 recordsLinked to original sources

arXiv2021

Risk and return prediction for pricing portfolios of non-performing consumer credit

We design a system for risk-analyzing and pricing portfolios of non-performing consumer credit loans. The rapid development of credit lending business for consumers heightens the need for trading portfolios formed by overdue loans as a manner of risk transferring. However, the problem is nontrivial technically and related research is absent. We tackle the challenge by building a bottom-up architecture, in which we model the distribution of every single loan's repayment rate, followed by modeling the distribution of the portfolio's overall repayment rate. To address the technical issues encountered, we adopt the approaches of simultaneous quantile regression, R-copula, and Gaussian one-factor copula model. To our best knowledge, this is the first study that successfully adopts a bottom-up system for analyzing credit portfolio risks of consumer loans. We conduct experiments on a vast amount of data and prove that our methodology can be applied successfully in real business tasks.

q-fin.RM↗

arXiv2020

Memory-Gated Recurrent Networks

The essence of multivariate sequential learning is all about how to extract dependencies in data. These data sets, such as hourly medical records in intensive care units and multi-frequency phonetic time series, often time exhibit not only strong serial dependencies in the individual components (the "marginal" memory) but also non-negligible memories in the cross-sectional dependencies (the "joint" memory). Because of the multivariate complexity in the evolution of the joint distribution that underlies the data generating process, we take a data-driven approach and construct a novel recurrent network architecture, termed Memory-Gated Recurrent Networks (mGRN), with gates explicitly regulating two distinct types of memories: the marginal memory and the joint memory. Through a combination of comprehensive simulation studies and empirical experiments on a range of public datasets, we show that our proposed mGRN architecture consistently outperforms state-of-the-art architectures targeting multivariate time series.

cs.LG↗

arXiv2020

The Causal Learning of Retail Delinquency

This paper focuses on the expected difference in borrower's repayment when there is a change in the lender's credit decisions. Classical estimators overlook the confounding effects and hence the estimation error can be magnificent. As such, we propose another approach to construct the estimators such that the error can be greatly reduced. The proposed estimators are shown to be unbiased, consistent, and robust through a combination of theoretical analysis and numerical testing. Moreover, we compare the power of estimating the causal quantities between the classical estimators and the proposed estimators. The comparison is tested across a wide range of models, including linear regression models, tree-based models, and neural network-based models, under different simulated datasets that exhibit different levels of causality, different degrees of nonlinearity, and different distributional properties. Most importantly, we apply our approaches to a large observational dataset provided by a global technology firm that operates in both the e-commerce and the lending business. We find that the relative reduction of estimation error is strikingly substantial if the causal effects are accounted for correctly.

q-fin.RM↗

arXiv2020

Understanding Distributional Ambiguity via Non-robust Chance Constraint

This paper provides a non-robust interpretation of the distributionally robust optimization (DRO) problem by relating the distributional uncertainties to the chance probabilities. Our analysis allows a decision-maker to interpret the size of the ambiguity set, which is often lack of business meaning, through the chance parameters constraining the objective function. We first show that, for general $ϕ$-divergences, a DRO problem is asymptotically equivalent to a class of mean-deviation problems. These mean-deviation problems are not subject to uncertain distributions, and the ambiguity radius in the original DRO problem now plays the role of controlling the risk preference of the decision-maker. We then demonstrate that a DRO problem can be cast as a chance-constrained optimization (CCO) problem when a boundedness constraint is added to the decision variables. Without the boundedness constraint, the CCO problem is shown to perform uniformly better than the DRO problem, irrespective of the radius of the ambiguity set, the choice of the divergence measure, or the tail heaviness of the center distribution. Thanks to our high-order expansion result, a notable feature of our analysis is that it applies to divergence measures that accommodate well heavy tail distributions such as the student $t$-distribution and the lognormal distribution, besides the widely-used Kullback-Leibler (KL) divergence, which requires the distribution of the objective function to be exponentially bounded. Using the portfolio selection problem as an example, our comprehensive testings on multivariate heavy-tail datasets, both synthetic and real-world, shows that this business-interpretation approach is indeed useful and insightful.

math.OC↗

arXiv2012

SDSS quasars in the WISE preliminary data release and quasar candidate selection with optical/infrared colors

We present a catalog of 37,842 quasars in the SDSS Data Release 7, which have counterparts within 6" in the WISE Preliminary Data Release. The overall WISE detection rate of the SDSS quasars is 86.7%, and it decreases to less than 50.0% when the quasar magnitude is fainter than $i=20.5$. We derive the median color-redshift relations based on this SDSS-WISE quasar sample and apply them to estimate the photometric redshifts of the SDSS-WISE quasars. We find that by adding the WISE W1- and W2-band data to the SDSS photometry we can increase the photometric redshift reliability, defined as the percentage of sources with the photometric and spectroscopic redshift difference less than 0.2, from 70.3% to 77.2%. We also obtain the samples of WISE-detected normal and late-type stars with SDSS spectroscopy, and present a criterion in the $z-W1$ versus $g-z$ color-color diagram, $z-W1>0.66(g-z)+2.01$, to separate quasars from stars. With this criterion we can recover 98.6% of 3089 radio-detected SDSS-WISE quasars with redshifts less than four and overcome the difficulty in selecting quasars with redshifts between 2.2 and 3 from SDSS photometric data alone. We also suggest another criterion involving the WISE color only, $W1-W2>0.57$, to efficiently separate quasars with redshifts less than 3.2 from stars. In addition, we compile a catalog of 5614 SDSS quasars detected by both WISE and UKIDSS surveys and present their color-redshift relations in the optical and infrared bands. By using the SDSS $ugriz$, UKIDSS YJHK and WISE W1- and W2-band photometric data, we can efficiently select quasar candidates and increase the photometric redshift reliability up to 87.0%. We discuss the implications of our results on the future quasar surveys. An updated SDSS-WISE quasar catalog consisting of 101,853 quasars with the recently released WISE all-sky data is also provided.

astro-ph.CO↗

arXiv2012

Selecting Quasar Candidates by a SVM Classification System

We develop and demonstrate a classification system constituted by several Support Vector Machines (SVM) classifiers, which can be applied to select quasar candidates from large sky survey projects, such as SDSS, UKIDSS, GALEX. How to construct this SVM classification system is presented in detail. When the SVM classification system works on the test set to predict quasar candidates, it acquires the efficiency of 93.21% and the completeness of 97.49%. In order to further prove the reliability and feasibility of this system, two chunks are randomly chosen to compare its performance with that of the XDQSO method used for SDSS-III's BOSS. The experimental results show that the high faction of overlap exists between the quasar candidates selected by this system and those extracted by the XDQSO technique in the dereddened i-band magnitude range between 17.75 and 22.45, especially in the interval of dereddened i-band magnitude < 20.0. In the two test areas, 57.38% and 87.15% of the quasar candidates predicted by the system are also targeted by the XDQSO method. Similarly, the prediction of subcategories of quasars according to redshift achieves a high level of overlap with these two approaches. Depending on the effectiveness of this system, the SVM classification system can be used to create the input catalog of quasars for the GuoShouJing Telescope (LAMOST) or other spectroscopic sky survey projects. In order to get higher confidence of quasar candidates, cross-result from the candidates selected by this SVM system with that by XDQSO method is applicable.

astro-ph.IM↗