基于语义的孤立点检测

目前,大多数孤立点检测算法仅仅考虑了数据集本身,而没有考虑数据集所蕴涵的语义知识。本文我们通过分析隐藏在Web日志中的语义知识来进行孤立点检测,提出了一种基于语义的孤立点挖掘方法。该方法基于Web日志中记录的各个项满足的数值关系来分析其中隐含的语义信息,并根据这些语义信息的重要性给出一个综合衡量其相关性的指标。实验结果表明,该方法是可行的、有效的。     

An outlier test for linear processes — II. Large contamination

In Flak/Schmid (1993) an outlier test for linear processes was introduced. The test statistic bases on a comparison of each observation with a one-step predictor. It was assumed that an upper bound for the total number of outlierss _n is known, wheren denotes the sample size. The asymptotic distribution of the test statistic was derived under the assumption thats _n/n → 0 ands _n → ∞ asn → ∞. This note deals with the asymptotic behaviour of this quantity, ifs _n/n →p ₀ ∈ (0, 1).     

Analysis on structural changes in the macroeconomic data series with the empirical evidence from China

Analysis on structural changes in macroeconomic data series has been the key issue for studying data quality. This paper studies the structural changes in China’s 36 macroeconomic time series using joint estimation model, and we find out the characteristics and movement pattern for the outliers. Our results show that most outliers show up more or less in groups, indicating that there is a significant correlation between them. The isolated outliers are not the main characteristic of China’s macroeconomic time series. Nearly all the original series contain the obvious skewness and kurtosis; hence, the hypothesis of normality is significantly rejected. Most original and outlier correction series show the non-autoregressive conditional heteroskedasticity (ARCH) characteristic, but the p value for ARCH2, ARCH4, and ARCH8 is very different. __________ Translated from Economic Research Journal (经济研究), 2005, (1) (in Chinese)     

Technical Efficiency of the Brazilian Municipalities: Correcting Nonparametric Frontier Measurements for Outliers

In this paper we estimate the DEA technical efficiency for 4796 Brazilian municipalities, by applying a recently proposed “Jackstrap” method, which combines Bootstrap and Jackknife resampling techniques, to reduce the effect of outliers and possible errors in the data set. We perform calculations to identify and eliminate high leverage municipalities, using different variants of Data Envelopment Analysis (DEA), as well as Free Disposal Hull (FDH). Corroborating previous results, efficiency results for the Brazilian municipalities show a clear relationship between the size of the municipality and its efficiency scores. Indeed, under both DEA variants, smaller cities tend to be less efficient than larger ones hence indicating that the quality of the frontier adjustment improves significantly as the size of the municipality increases. We present arguments that may explain to some extent these findings, such as economies of scale and the excess spending due to revenue from royalties. However, such effects require further, more careful examination.     

Wages and Wage Inequality in South Africa 1994–2011: Part 1 – Wage Measurement and Trends

We analyse the long‐term trends in wage inequality in South Africa, using household survey data. We show that the trends in household income inequality are largely driven by changes in wage inequality. Given the detailed nature of our series we show that measurement issues and breaks in the series need to be dealt with in order to draw robust conclusions from the data. Most standard inequality measures show that wage inequality has increased over the period. Nevertheless the choice of measure matters, because there are different trends in different parts of the distribution. It appears that the distribution below the median has become more compressed, while the top of the wage distribution has moved away from the median. The inequality in the labour market translates into even higher inequality in society given that high earners tend to live together with other high earners while low wage individuals often end up sharing their incomes with the unemployed. Furthermore there are many South Africans with access to no wage income. Given the trends analysed here it is not surprising that overall inequality in South Africa has not come down or has even increased since the end of apartheid.     

FURTHER EVIDENCE ON MEAN REVERSION IN THE AUSTRALIAN EXCHANGE RATE

In this paper we re‐examine the purchasing power parity (PPP) in Australia for the period February 1970 to April 2005 using an alternative method relative to the previous studies. We underlined large shocks due to depreciations that affected the Australian exchange rate, using outlier methodology. Once we adjusted the data of these outliers that had large, but either temporary or permanent effects on the series, our results show that there is no tendency for PPP in Australia to hold in the long run during this period.     

The core and periphery of the world economy

This paper reviews three models of foreign trade, including the Heckscher-Ohlin model, the new trade theory based on increasing returns to scale, and the model of economic geography and trade with agglomeration effects. It demonstrates that gravity models perform relatively well for differentiated and non-differentiated products. This result supports Hummels' and Levinsohn's () critique of the new theory foundation of the gravity equation. Furthermore, the bilateral trade relations of peripheral countries are often identified as outliers. This pattern of outliers is consistent with the model of geography and trade.     

Exploring promising vacant technology areas in a technology-oriented company based on bibliometric analysis and visualisation*

Under technology convergence phenomenon and fast-changing social demands, it is inevitable to collect and analyze a large amount of information in order to determine a vacant technology area that can preempt the market. Thus, this research aims to propose a systematic approach to identifying vacant technology areas by reflecting the technological competitiveness of company. Vacant technology is identified as underdeveloped and undeveloped vacant areas for a firm in both macroscopic and microscopic view by generating a technology level map in a practical view. In addition, vacant technology can be identified as undiscovered vacant areas in a microscopic view by using generative topology mapping (GTM) and local outlier factor (LOF). In a respective view, the promising vacant areas are defined by developing various indicators that represent competitive strength, growth stage, and marketability in both macro and micro level, and novelty indicators that represent originality and marketability in the micro level. The proposed approach is applied to the information & communication technology (ICT) R&D technology classification to validate its usefulness. The results can be used to enhance the practical application of R&D technology planning, understand the current technology level, and perform future technology planning.     

An outlier test for linear processes

In this paper we introduce an outlier test for linear processes. It is assumed that an upper bound for the number of outliers is known which is not too big in relation to the sample size. The test statistic bases on the comparison of the observations with certain predictors. We discuss the asymptotical behaviour of the test statistic under the null hypothesis ‘no outlier’ and derive the asymptotic distribution for the case that the distribution of the squared white noise process belongs to a certain subset of the domain of attraction of the Gumbel distribution. Especially the most important case in applications, the Gaussian white noise is included.     

数据流挖掘及其在持续审计中的可用性研究

随着企业信息化程度的提高和互联网的普及,每天都会产生海量的实时数据,而数据流挖掘则为分析海量数据提供了一种新途径。数据流挖掘中的聚类、分类、离群点检测等算法的研究取得了进展,为在持续审计中应用数据流挖掘提供了可行性。本文提出的一种基于数据流挖掘的持续审计模型,克服了传统持续审计模型对审计端的存储能力要求高、占用大量硬件资源、联机分析时间长、对异常数据的发现滞后等缺点。