Pub Date : 2020-10-14DOI: 10.1285/I20705948V13N2P293
B. Oluyede, H. Jimoh, D. Wanduku, B. Makubate
We introduce a new distribution via the Marshall-Olkin generator called the Marshall-Olkin Log-logistic Erlang-Truncated Exponential (MOLLoGETE) distribution. Some structural properties of the distribution including series expansion of the density function, sub-models, hazard function, moments, conditional moments, mean deviations, distribution of order statistics, R´enyi entropy and maximum likelihood estimates are presented. The new density function is an infinite linear combinations of Burr XII-Erlang-Truncated Exponential distributions. The new generalization is applied to real data sets to evaluate the model performance.
我们通过Marshall-Olkin生成器引入了一种新的分布，称为Marshall-Olkin-Log-listic-Erlang截断指数（MOLLoGETE）分布。给出了分布的一些结构性质，包括密度函数的级数展开、子模型、危险函数、矩、条件矩、平均偏差、阶统计量的分布、R´enyi熵和最大似然估计。新的密度函数是Burr XII Erlang截断指数分布的无限线性组合。将新的泛化方法应用于实际数据集，以评估模型的性能。
{"title":"A New Generalized Log-Logistic Erlang Truncated Exponential Distribution with Applications","authors":"B. Oluyede, H. Jimoh, D. Wanduku, B. Makubate","doi":"10.1285/I20705948V13N2P293","DOIUrl":"https://doi.org/10.1285/I20705948V13N2P293","url":null,"abstract":"We introduce a new distribution via the Marshall-Olkin generator called the Marshall-Olkin Log-logistic Erlang-Truncated Exponential (MOLLoGETE) distribution. Some structural properties of the distribution including series expansion of the density function, sub-models, hazard function, moments, conditional moments, mean deviations, distribution of order statistics, R´enyi entropy and maximum likelihood estimates are presented. The new density function is an infinite linear combinations of Burr XII-Erlang-Truncated Exponential distributions. The new generalization is applied to real data sets to evaluate the model performance.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"293-349"},"PeriodicalIF":0.7,"publicationDate":"2020-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1285/I20705948V13N2P293","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41512778","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-14DOI: 10.1285/I20705948V13N2P375
Farideh Tavazoee, D. Buscaldi, F. Mola, C. Conversano
The role of Twitter as a platform to share opinions has been growing in the recent years especially since it has been widely used by public personae such as politicians, personalities of the show business, and other influencers to communicate with the public. For these reasons, the use of social bots to manipulate information and influence people's opinions is also growing. In this paper, we use a supervised classification model to distinguish bots from legitimate users on Twitter. More specifically, we show the importance of sentiment features in bot-human account detection. Moreover, we evaluate our detection model by testing on Russian bot accounts who are the most recent set of social bots that appeared on Twitter to show that these techniques may be easily adapted to work on new, unseen types of social bots.
{"title":"Empowering Detection of Malicious Social Bots and Content Spammers on Twitter by Sentiment Analysis","authors":"Farideh Tavazoee, D. Buscaldi, F. Mola, C. Conversano","doi":"10.1285/I20705948V13N2P375","DOIUrl":"https://doi.org/10.1285/I20705948V13N2P375","url":null,"abstract":"The role of Twitter as a platform to share opinions has been growing in the recent years especially since it has been widely used by public personae such as politicians, personalities of the show business, and other influencers to communicate with the public. For these reasons, the use of social bots to manipulate information and influence people's opinions is also growing. In this paper, we use a supervised classification model to distinguish bots from legitimate users on Twitter. More specifically, we show the importance of sentiment features in bot-human account detection. Moreover, we evaluate our detection model by testing on Russian bot accounts who are the most recent set of social bots that appeared on Twitter to show that these techniques may be easily adapted to work on new, unseen types of social bots.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"375-389"},"PeriodicalIF":0.7,"publicationDate":"2020-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43287275","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-10-14DOI: 10.1285/I20705948V13N2P284
A. Olosunde, A. T. Soyinkab
It is common to assume a normal distribution when discriminating and classifying a multivariate data based on some attributes. But when such data is lighter or heavier in both tails than the normal distribution, then the probability of misclassification becomes higher giving unreliable result. This study proposed multivariate exponential power distribution a family of elliptically contoured model as underlining model for discrimination and classification. The distribution has a shape parameter which regulate the tail of the symmetric distribution to mitigate the problem of both lighter and heavier tails data, this generalizes the normal distribution and thus will definitely gives a lower misclassification error in discrimination and classification. The resulting discriminant model was compared with fisher linear discriminant function when applying to real data.
{"title":"Discrimination and Classification model from Multivariate Exponential Power Distribution","authors":"A. Olosunde, A. T. Soyinkab","doi":"10.1285/I20705948V13N2P284","DOIUrl":"https://doi.org/10.1285/I20705948V13N2P284","url":null,"abstract":"It is common to assume a normal distribution when discriminating and classifying a multivariate data based on some attributes. But when such data is lighter or heavier in both tails than the normal distribution, then the probability of misclassification becomes higher giving unreliable result. This study proposed multivariate exponential power distribution a family of elliptically contoured model as underlining model for discrimination and classification. The distribution has a shape parameter which regulate the tail of the symmetric distribution to mitigate the problem of both lighter and heavier tails data, this generalizes the normal distribution and thus will definitely gives a lower misclassification error in discrimination and classification. The resulting discriminant model was compared with fisher linear discriminant function when applying to real data.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"284-292"},"PeriodicalIF":0.7,"publicationDate":"2020-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45164898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-02-05DOI: 10.1285/I20705948V13N1P256
Giovanna Di Lorenzo, A. Orlando, Massimo Politano
The reverse mortgage market has been expanding rapidly in developed economies in recent years. Reverse mortgages provide an alternative source of funding for retirement income and health care costs. Increase in life expectancies and decrease in the real income at retirement continue to worry the those who are retired or close to retirement. Therefore, financial products that help to alleviate the “risk of living longer” continue to be attractive among the retirees. Reverse mortgage contracts involve a range of risks from the insurer’s perspective. When the outstanding balance exceeds the housing value before the loan is settled, the insurer suffers an exposure to crossover risk induced by three risk factors: interest rates, house prices and mortality rates. We analyse the combined impact of these risks on the pricing and the risk profile of reverse mortgage loans in a stochastic interest-mortality-house pricing model. Our results show shows that pricing of reverse mortgages loans does not accurately assess the risks underwritten by reverse mortgages lenders and that failing to take into account mortality improvements substantially underestimates the longevity risk involved in reverse mortgage loans.
{"title":"The security mortgage valuation in a stochastic perspective","authors":"Giovanna Di Lorenzo, A. Orlando, Massimo Politano","doi":"10.1285/I20705948V13N1P256","DOIUrl":"https://doi.org/10.1285/I20705948V13N1P256","url":null,"abstract":"The reverse mortgage market has been expanding rapidly in developed economies in recent years. Reverse mortgages provide an alternative source of funding for retirement income and health care costs. Increase in life expectancies and decrease in the real income at retirement continue to worry the those who are retired or close to retirement. Therefore, financial products that help to alleviate the “risk of living longer” continue to be attractive among the retirees. Reverse mortgage contracts involve a range of risks from the insurer’s perspective. When the outstanding balance exceeds the housing value before the loan is settled, the insurer suffers an exposure to crossover risk induced by three risk factors: interest rates, house prices and mortality rates. We analyse the combined impact of these risks on the pricing and the risk profile of reverse mortgage loans in a stochastic interest-mortality-house pricing model. Our results show shows that pricing of reverse mortgages loans does not accurately assess the risks underwritten by reverse mortgages lenders and that failing to take into account mortality improvements substantially underestimates the longevity risk involved in reverse mortgage loans.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"256-267"},"PeriodicalIF":0.7,"publicationDate":"2020-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48259100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-02-05DOI: 10.1285/I20705948V13N1P146
P. N. Inverardi, E. Taufer
The paper investigates the use of a finite mixture model with an additional uniform density for outlier detection and robust estimation. The main contribution of this paper lies in the analysis of the properties of the improper component and the introduction of a modified EM algorithm which, beyond providing the maximum likelihood estimates of the mixture parameters, endogenously provides a numerical value for the density of the uniform distribution used for the improper component. The mixing proportion of outliers may be known or unknown. Applications to robust estimation and outlier detection will be discussed with particular attention to the normal mixture case.
{"title":"Outlier detection through mixtures with an improper component","authors":"P. N. Inverardi, E. Taufer","doi":"10.1285/I20705948V13N1P146","DOIUrl":"https://doi.org/10.1285/I20705948V13N1P146","url":null,"abstract":"The paper investigates the use of a finite mixture model with an additional uniform density for outlier detection and robust estimation. The main contribution of this paper lies in the analysis of the properties of the improper component and the introduction of a modified EM algorithm which, beyond providing the maximum likelihood estimates of the mixture parameters, endogenously provides a numerical value for the density of the uniform distribution used for the improper component. The mixing proportion of outliers may be known or unknown. Applications to robust estimation and outlier detection will be discussed with particular attention to the normal mixture case.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"146-163"},"PeriodicalIF":0.7,"publicationDate":"2020-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"44548209","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-02-05DOI: 10.1285/I20705948V13N1P1
Norihiro Mita, H. Sasaki, K. Kani, A. Tabuchi, H. Hara
To establish the computer assisted system of the visual acuity test, we propose a statistical modelling of the visual acuity measurement and its multiple test procedure. The psychometric functions for individual patients are produced by the logistic regression combined with the guessing rate. We adopt test statistics based on (i) psychometric functions (the Cochran-Mantel-Haenszel method) and (ii) psychophysical thresholds (the delta method). The multiple comparisons are performed by the step-down procedure with Ryan-Einot-Gabriel-Welsch (REGW) significance levels. To show the practical effectiveness of our system, we present a numerical example of four patient groups.
为了建立计算机辅助视力测试系统，我们提出了一种视力测量的统计模型及其多重测试程序。个体患者的心理测量函数是通过逻辑回归与猜测率相结合产生的。我们采用基于（i）心理测量函数（Cochran-Mantel-Haenszel方法）和（ii）心理物理阈值（delta方法）的测试统计。多重比较通过降压程序与Ryan Einot Gabriel Welsch（REGW）显著性水平进行。为了显示我们的系统的实际有效性，我们给出了四个患者组的数值示例。
{"title":"A statistical modelling of the visual acuity measurement and its multiple test procedure","authors":"Norihiro Mita, H. Sasaki, K. Kani, A. Tabuchi, H. Hara","doi":"10.1285/I20705948V13N1P1","DOIUrl":"https://doi.org/10.1285/I20705948V13N1P1","url":null,"abstract":"To establish the computer assisted system of the visual acuity test, we propose a statistical modelling of the visual acuity measurement and its multiple test procedure. The psychometric functions for individual patients are produced by the logistic regression combined with the guessing rate. We adopt test statistics based on (i) psychometric functions (the Cochran-Mantel-Haenszel method) and (ii) psychophysical thresholds (the delta method). The multiple comparisons are performed by the step-down procedure with Ryan-Einot-Gabriel-Welsch (REGW) significance levels. To show the practical effectiveness of our system, we present a numerical example of four patient groups.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"1-15"},"PeriodicalIF":0.7,"publicationDate":"2020-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1285/I20705948V13N1P1","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43221644","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-02-05DOI: 10.1285/I20705948V13N1P16
J. Divecha, Jigneshkumar Gondaliya
The assumption of carryover effects is unavoidable due to the very nature of crossover designs. Even in case of crossover design with washout period, the hypothesis of no carryover effect should be tested and established. On the other hand, this assumption makes the analysis difficult and potentially biased or inefficient in case of two treatment two period crossover design. For a reasonable estimation, experimenters are advocated to employ a two period three treatment crossover designs, or a three period two treatment crossover design. In this article, we present optional analyses of a uniform three period three treatment crossover design, consisting of a placebo and two active treatments. We develop a test for detecting presence of carryover effects which directs experimenter for a proper analysis of his crossover trial. We present ANOVA for each of the three possible carryover models, that both, single, or none of the active treatments has carryover effect, and illustrate through an example.
{"title":"Optional analyses of crossover trials having two treatments and a placebo","authors":"J. Divecha, Jigneshkumar Gondaliya","doi":"10.1285/I20705948V13N1P16","DOIUrl":"https://doi.org/10.1285/I20705948V13N1P16","url":null,"abstract":"The assumption of carryover effects is unavoidable due to the very nature of crossover designs. Even in case of crossover design with washout period, the hypothesis of no carryover effect should be tested and established. On the other hand, this assumption makes the analysis difficult and potentially biased or inefficient in case of two treatment two period crossover design. For a reasonable estimation, experimenters are advocated to employ a two period three treatment crossover designs, or a three period two treatment crossover design. In this article, we present optional analyses of a uniform three period three treatment crossover design, consisting of a placebo and two active treatments. We develop a test for detecting presence of carryover effects which directs experimenter for a proper analysis of his crossover trial. We present ANOVA for each of the three possible carryover models, that both, single, or none of the active treatments has carryover effect, and illustrate through an example.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"16-30"},"PeriodicalIF":0.7,"publicationDate":"2020-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1285/I20705948V13N1P16","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43658692","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-02-05DOI: 10.1285/I20705948V13N1P75
H. Kuswanto, Nesia Balqis, H. Ohwada, S.F. Toha
Colorectal cancer has become one of the cancer types with high incidence rate all over the world. Various eorts have been carried out to nd a way to decrease the risk of cancer. Chemotherapy using 5-Fluorouracil (5-FU) is one of the common cancer treatments that is expected to drive the white blood cell (WBC) level into the normal level. This research investigates the factors inuencing the change of WBC level in cancer patients treated with 5-FU combined with physical treatment in the form of footsteps. By focusing on the change of WBC level, i.e. decreasing or increasing the WBC level as the response, probit regression was applied to the data measured from 28 cancer patients who have undergone 14 days of treatment. The probit regression found that age of the patient, average number of daily footsteps and the dose of 5-FU signcantly inuence the change of WBC. The regression is able to classify the case with a satisfactory results, i.e. 85.71% classication accuracy. This nding can be a guideline to better treat the colorectal cancer patient to reach a normal WBC.
{"title":"Modelling the Change of White Blood Cell on Colorectal Cancer Treatment Using Probit Regression","authors":"H. Kuswanto, Nesia Balqis, H. Ohwada, S.F. Toha","doi":"10.1285/I20705948V13N1P75","DOIUrl":"https://doi.org/10.1285/I20705948V13N1P75","url":null,"abstract":"Colorectal cancer has become one of the cancer types with high incidence rate all over the world. Various eorts have been carried out to nd a way to decrease the risk of cancer. Chemotherapy using 5-Fluorouracil (5-FU) is one of the common cancer treatments that is expected to drive the white blood cell (WBC) level into the normal level. This research investigates the factors inuencing the change of WBC level in cancer patients treated with 5-FU combined with physical treatment in the form of footsteps. By focusing on the change of WBC level, i.e. decreasing or increasing the WBC level as the response, probit regression was applied to the data measured from 28 cancer patients who have undergone 14 days of treatment. The probit regression found that age of the patient, average number of daily footsteps and the dose of 5-FU signcantly inuence the change of WBC. The regression is able to classify the case with a satisfactory results, i.e. 85.71% classication accuracy. This nding can be a guideline to better treat the colorectal cancer patient to reach a normal WBC.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"75-85"},"PeriodicalIF":0.7,"publicationDate":"2020-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47812476","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-02-05DOI: 10.1285/I20705948V13N1P47
T. Ogura, H. Murakami
Canonical correlation analysis (CCA) is often used to analyze correlations between the variables of two random vectors. As an extension of CCA, multiple-set canonical correlation analysis (MCCA) was proposed to analyze correlations between multiple-set random vectors. However, sometimes interpreting MCCA results may not be as straightforward as interpreting CCA results. Principal CCA (PCCA), which uses CCA between two sets of principal component (PC) scores, was proposed to address these difficulties in CCA. We propose multiple-set PCCA (MPCCA) by applying the idea to multiple-set of PC scores. PCs are ranked in descending order according to the amount of information they contain. Therefore, it is enough to use only a few PC scores from the top instead of using all PC scores. Decreasing the number of PC makes it easy to interpret the result. We confirmed the effectiveness of MPCCA using simulation studies and a practical example.
{"title":"Canonical Correlation Analysis of Principal Component Scores for Multiple-set Random Vectors","authors":"T. Ogura, H. Murakami","doi":"10.1285/I20705948V13N1P47","DOIUrl":"https://doi.org/10.1285/I20705948V13N1P47","url":null,"abstract":"Canonical correlation analysis (CCA) is often used to analyze correlations between the variables of two random vectors. As an extension of CCA, multiple-set canonical correlation analysis (MCCA) was proposed to analyze correlations between multiple-set random vectors. However, sometimes interpreting MCCA results may not be as straightforward as interpreting CCA results. Principal CCA (PCCA), which uses CCA between two sets of principal component (PC) scores, was proposed to address these difficulties in CCA. We propose multiple-set PCCA (MPCCA) by applying the idea to multiple-set of PC scores. PCs are ranked in descending order according to the amount of information they contain. Therefore, it is enough to use only a few PC scores from the top instead of using all PC scores. Decreasing the number of PC makes it easy to interpret the result. We confirmed the effectiveness of MPCCA using simulation studies and a practical example.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"47-74"},"PeriodicalIF":0.7,"publicationDate":"2020-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49330884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-02-05DOI: 10.1285/I20705948V13N1P211
C. Caroni, F. Pierri
We examine the time until closure of Small Business Enterprises in Umbria, Italy between 2008 and 2013, and the factors that influence it. Earlier analysis, using Cox regression, considered failure (closure) from any cause. However, there are different reasons for inactivity: voluntary winding-up (1808 of 15184 firms in our data, 59.3% of the 3049 failures); bankruptcy (236, 7.7%); and closure without action by creditors or courts (1005, 33.0%). While the earlier analysis provides a valuable overall picture, it is also interesting to ex- amine the separate causes, their rates of occurrence and which factors influence them separately. We do this using competing risks analyses, employing both of the regression methods that are prominent in the literature, based on cause-specific and sub-distribution hazard functions (Fine-Gray model). Furthermore, a proportional odds model was used to estimate cumulative incidences of failure by cause. Data included the firm's year of foundation, location, legal form and sector of activity. Financial indexes were constructed from annual balance sheets. The date and reason for closure were recorded if the firrm ceased activity. Findings included major differences between types of firm; for example, cooperatives had greatly increased hazards for winding-up (HR of 2.44 and 2.61 in the two approaches) but greatly reduced hazards for closure (0.48 and 0.45) compared to publicly traded companies. All-causes analysis averaged these strong effects into an insignicant one (1.05). Coefficients from the proportional odds model were similar to those from the Fine-Gray model, but have the advantage of interpretability.
{"title":"Different causes of closure of Small Business Enterprises: alternative models for competing risks survival analysis","authors":"C. Caroni, F. Pierri","doi":"10.1285/I20705948V13N1P211","DOIUrl":"https://doi.org/10.1285/I20705948V13N1P211","url":null,"abstract":"We examine the time until closure of Small Business Enterprises in Umbria, Italy between 2008 and 2013, and the factors that influence it. Earlier analysis, using Cox regression, considered failure (closure) from any cause. However, there are different reasons for inactivity: voluntary winding-up (1808 of 15184 firms in our data, 59.3% of the 3049 failures); bankruptcy (236, 7.7%); and closure without action by creditors or courts (1005, 33.0%). While the earlier analysis provides a valuable overall picture, it is also interesting to ex- amine the separate causes, their rates of occurrence and which factors influence them separately. We do this using competing risks analyses, employing both of the regression methods that are prominent in the literature, based on cause-specific and sub-distribution hazard functions (Fine-Gray model). Furthermore, a proportional odds model was used to estimate cumulative incidences of failure by cause. Data included the firm's year of foundation, location, legal form and sector of activity. Financial indexes were constructed from annual balance sheets. The date and reason for closure were recorded if the firrm ceased activity. Findings included major differences between types of firm; for example, cooperatives had greatly increased hazards for winding-up (HR of 2.44 and 2.61 in the two approaches) but greatly reduced hazards for closure (0.48 and 0.45) compared to publicly traded companies. All-causes analysis averaged these strong effects into an insignicant one (1.05). Coefficients from the proportional odds model were similar to those from the Fine-Gray model, but have the advantage of interpretability.","PeriodicalId":44770,"journal":{"name":"Electronic Journal of Applied Statistical Analysis","volume":"13 1","pages":"211-228"},"PeriodicalIF":0.7,"publicationDate":"2020-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43112283","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：