Xiaohui Chang

Social commerce and location-based services provide a data platform for coexisting and competing businesses in geographical neighborhoods. Our research is aimed at mining data from such platforms to gain valuable insights for better support to strategic and operational business decisions. We develop a computational framework for predicting business performance that takes into account both intrinsic (e.g., attributes) and extrinsic (e.g., competitions) factors. Our experiments on synthetic and real datasets demonstrated superiority of a hybrid prediction model that adopts both link-based and context-based assumptions.

Variogram estimation plays a vastly important role in spatial modeling. Different methods
for variogram estimation can be largely classified into least squares methods and likelihood
based methods. A general framework to estimate the variogram through a set of estimating
equations is proposed. This approach serves as an alternative approach to likelihood based
methods and includes commonly used least squares approaches as its special cases. The
proposed method is highly efficient as a low dimensional representation of the weight
matrix is employed. The statistical efficiency of various estimators is explored and the lag
effect is examined. An application to a hydrology data set is also presented.

Based on daily and one-minute high-frequency returns, this paper examines the
lead-lag dependence between the CSI 300 index spot and futures markets from 2010 to 2014. The
nonparametric and nonlinear thermal optimal path method is adopted. Empirical results of the
daily data indicate that the lead-lag relationship between the two markets is within one day but
this relationship is volatile since neither of the two possible situations (the futures leads or lags
behind the spot market) takes a dominant place. Besides, our results from high-frequency data
demonstrate that there is a price discovery in the Chinese futures market: the intraday one-minute
futures return leads the cash return by 0~5 minutes regardless of the price trend of the market.

Purpose – The purpose of this paper is to identify the bull and bear regimes in Chinese stock market and empirically analyze the dynamic relation of Chinese stock price-volume pre- and post- the Split Share Structure Reform.

Design/methodology/approach – The authors investigate the price-volume relationship in the Chinese stock market before and after the Split Share Structure Reform using Shanghai Composite Index daily data from July 1994 to April 2013. Using a two-state Markov-switching autoregressive model and a modified two-state Markov-switching vector autoregression model, this study identifies bull or bear market and also examine the existence of regime-dependent Granger causality.

Findings – Using a two-state Markov-switching autoregressive model, the authors detect structural changes in the market volatility due to the reform, and find evidence of a positive rather than an asymmetric price-volume contemporaneous correlation. There is a strong dynamic Granger causal relation from stock returns to trading volume before and after the reform regardless of the market conditions, but the causal effects of volume on returns are only seen in the bear markets before the reform. The model is robust when using different stock indices and time periods.

Originality/value – The work is different from previous studies in the following aspects: most of the existing empirical literature focus on the well-developed economies, but our interest lies in the emerging Chinese market that has witnessed rapid growth in the past decade; in contrast to many works in the literature that examine the price-volume relationship during one market condition, the authors compare the relationship in a bull market with that in a bear market, using a two-state MS-AR model; the authors also employ a modified two-state Markov-switching vector autoregression model to examine the existence of regime-dependent Granger causality; as the most massive systematic reform for the Chinese stock market since its inception in 2005, the Split Share Structure Reform has a profound impact on the Chinese stock market, thus it is of vital importance to explore its effects on both the price-volume relationship and the market structure.

We propose a novel statistical framework by supplementing case-control data with summary statistics on the population at risk for a subset of risk factors. Our approach is to first form two unbiased estimating equations, one based on the case-control data and the other on both the case data and the summary statistics, and then optimally combine them to derive another estimating equation to be used for the estimation. The proposed method is computationally simple and more efficient than standard approaches based on case-control data alone. We also establish asymptotic properties of the resulting estimator, and investigate its finite-sample performance through simulation. As a substantive application, we apply the proposed method to investigate risk factors for endometrial cancer, by using data from a recently completed population-based case-control study and summary statistics from the Behavioral Risk Factor Surveillance System, the Population Estimates Program of the US Census Bureau, and the Connecticut Department of Transportation.

The location-scale and whitening properties of wavelets make them more favorable for interpolating high-frequency monitoring data than Fourier-based methods. In the past, wavelets have been used to simplify the dependence structure in multiple time or spatial series, but little has been done to apply wavelets as a modeling tool in a space–time setting, or, in particular, to take advantage of the localization of wavelets to capture the local dynamic characteristics of high-frequency meteorological data. This paper analyzes minute-by-minute atmospheric pressure data from the Atmospheric Radiation Measurement program using different wavelet coefficient structures at different scales and incorporating spatial structure into the model. This approach of modeling space–time processes using wavelets produces accurate point predictions with low uncertainty estimates, and also enables interpolation of available data from sparse monitoring stations to a high density grid and production of meteorological maps on large spatial and temporal scales.

Theoretical aspects of the decorrelation property of the discrete wavelet transform when applied to stochastic processes have been studied exclusively from the increasing-domain perspective, in which the distance between neighboring observations stays roughly constant as the number of observations increases. To understand the underlying data-generating process and to obtain good interpolations, fixed-domain asymptotics, in which the number of observations increases in a fixed region, is often more appropriate than increasing-domain asymptotics. In the fixed-domain setting, we prove that, for a general class of inhomogeneous covariance functions, with suitable choice of wavelet filters, the wavelet transform of a nonstationary process has mostly asymptotically uncorrelated components.