The Distribution Patterns of Valency-changing Verbs: An Approach of Quantitative Linguistics

The present study attempts to explore the distribution patterns of the valency-changing verbs from the perspective of quantitative linguistics. We took authentic spoken language data as the research materials. The corpus used in this paper is a self-built spoken English corpus containing about 21,000 words. We half-manually annotated the corpus with the help of SpaCy, a natural language processing tool. According to the annotation results and statistical data, we obtained a total of 217 valency-changing English verbs and 248 sentence components governed by them. After analysis, the current study came to the following conclusions: First, bivalent verbs are most frequent among the three types of valency-changing verbs; second, after fitting all the language data to different probability distributions, we found that the rank-frequency distributions of all the valency-changing English verbs with different numbers of obligatory arguments obey the power law, and the frequencies of bivalent valency-changing verbs obey other kinds of distributions such as the mixed Poisson distribution.

FAKTOR-FAKTOR YANG MEMPENGARUHI KASUS DBD DI SULAWESI SELATAN DENGAN MENGGUNAKAN REGRESI POISSON INVERSE GAUSSIAN

Poisson regression is used to model enumeration data such as data on the number of DHF cases. This model has the assumption that is fulfilled is the average and the variance must have the same value or it is called the equidispersion. But this assumption is not fulfilled because the data on the number of dengue cases experienced violations of this assumption. The violation is that the average value is smaller than the variance value or it is called overdispersion. This results in incorrect conclusions because the prediction standard error is underestimated. The way to prevent this is by combining the Poisson distribution and discrete or continuous distribution, this combination is called Mixed Poisson Distribution. Researchers use one of the Mixed Poisson methods, namely Inverse Gaussian Poisson Regression (PIG) because the method is used when the data is overdispersed and the parameters are known or close form on the likelihood function. Based on the results of the study, it is known that the height of the area is a factor that significantly influences DHF cases in South Sulawesi and the model form is as follows: π=exp(5,902-0,0004189 X_2)Keyword: DHF Cases; Poisson Regression; Overdispersion; Poisson Inverse Gaussian Regression;

FAKTOR-FAKTOR YANG MEMPENGARUHI KASUS DBD DI SULAWESI SELATAN DENGAN MENGGUNAKAN REGRESI POISSON INVERSE GAUSSIAN

Regresi Poisson digunakan untuk memodelkan data yang bersifat cacahan seperti data jumlah kasus DBD. Model ini memiliki asumsi yang dipenuhi ialah rata-rata dan variansinya harus memiliki nilai yang sama besar atau disebut equidispersi. Tapi asumsi tersebut tidak terpenuhi karena data jumlah kasus DBD mengalami pelanggaran Asumsi ini. Pelanggarannya ialah nilai rata-rata lebih kecil dari nilai variansi atau disebut overdispersi. Hal ini mengakibabkan kesimpulan yang diperoleh tidak benar karena pendugaan standar error mengalami underestimate. Cara untuk mencegahnya yaitu dengan menggabungkan antara distribusi poisson dan distribusi diskrit atau kontinu, penggabungan ini dinamakan Mixed Poisson Distribution. Peneliti menggunakan metode salah satu dari Mixed Poisson yaitu Regresi Poisson Inverse Gaussian (PIG) karena metode digunakan apabila data tersebut mengalami overdispersi dan parameter diketahui atau close form pada fungsi likelihood. Berdasarkan hasil dari penelitian diketahui bahwa ketinggian wilayah ialah faktor yang mempengaruhi kasus DBD di Sulawesi Selatan secara signifikan dan diperoleh bentuk model yaitu sebagai berikut:π=exp(5,902-0,0004189 X_2)Kata0Kunci : Kasus DBD;0Regresi Poisson;0Overdispersi;0Regresi Poisson0Inverse Gaussian;

Bivariate Mixed Poisson and Normal Generalised Linear Models with Sarmanov Dependence—An Application to Model Claim Frequency and Optimal Transformed Average Severity

The aim of this paper is to introduce dependence between the claim frequency and the average severity of a policyholder or of an insurance portfolio using a bivariate Sarmanov distribution, that allows to join variables of different types and with different distributions, thus being a good candidate for modeling the dependence between the two previously mentioned random variables. To model the claim frequency, a generalized linear model based on a mixed Poisson distribution -like for example, the Negative Binomial (NB), usually works. However, finding a distribution for the claim severity is not that easy. In practice, the Lognormal distribution fits well in many cases. Since the natural logarithm of a Lognormal variable is Normal distributed, this relation is generalised using the Box-Cox transformation to model the average claim severity. Therefore, we propose a bivariate Sarmanov model having as marginals a Negative Binomial and a Normal Generalized Linear Models (GLMs), also depending on the parameters of the Box-Cox transformation. We apply this model to the analysis of the frequency-severity bivariate distribution associated to a pay-as-you-drive motor insurance portfolio with explanatory telematic variables.