PURPOSES: This study evaluates the reliability of the patterns of changes in the road surface temperature during winter using a statistical technique. In addition, a flexible road segmentation method is developed based on the collected road surface temperature data. METHODS: To collect and analyze the data, a thermal mapping system that could be attached to a survey vehicle along with various other sensors was employed. We first selected the test route based on the date and the weather and topographical conditions, since these factors affect the patterns of changes in the road surface temperature. Each route was surveyed a total of 10 times on a round-trip basis at the same times (5 AM to 6 AM). A correlation analysis was performed to identify whether the weather conditions reported for the survey dates were consistent with the actual conditions. In addition, we developed a method for dividing the road into sections based on the consecutive changes in the road surface temperature for use in future applications. Specifically, in this method, the road surface temperature data collected using the thermal mapping system was compared continuously with the average values for the various road sections, and the road was divided into sections based on the temperature. RESULTS : The results showed that the comparison of the reported and actual weather conditions and the standard deviation in the observed road surface temperatures could produce a good indicator of the reliability of the patterns of the changes in the road surface temperature. CONCLUSIONS: This research shows how road surface temperature data can be evaluated using a statistical technique. It also confirms that roads should be segmented based on the changes in the temperature and not using a uniform segmentation method.

The paper contributes deriving the confidence intervals according to the use types of hypothesis tests. The guidelines of usuage on the types of hypothesis tests and interval estimation are proposed. These formulars can be used to evaluate the effects of quality improvement activity.

This paper presents various interval estimation methods of binomial proportion for small n in multi-product small volume production and extremely small ^P like PPM or PPB fraction of defectives. This study classifies interval estimation of binomial proportion into three categories such as exact, approximate, Bayesian methods. These confidence intervals proposed in this paper can be applied to attribute process capability and attribute acceptance sampling plan for PPM or PPB.

  ANOVA is widely used for measurement system analysis. It assumes that the measurement error is normally distributed, which may not be seen in certain industrial cases. In this study, the exact and bootstrap confidence intervals for precision-to-toleranc

This paper is to propose two computation procedures of reliability measures for large interval data. First method is efficient to verify the relationship among four reliability measures such as F(t), R(t), f(t) and λ(t). Another method is effective to interpret the concept of various reliability measures. This study is also to reinterpret and recompute the errors of four reliability measures discovered in the reliability textbooks. Various numerical examples are presented to illustrate the application of two proposed procedures.

We consider a mixed model with covariates considered as fixed effects and a random factor. In this paper, we consider methods for constructing confidence intervals on measures of variability in repeatability and reproducibility study to the mixed model with fixed effects and random effects. Computer simulation is used to determine how well confidence intervals maintain the stated confidence level and compare confidence interval lengths for the methods. A numerical example is considered to illustrate the confidence intervals proposed.

This paper is to propose tolerance intervals for expected time at the given reliability and confidence level for continuous and discrete reliability model. We consider guaranteed - coverage tolerance intervals, that is, reliability - confidence level tolerance intervals. These proposed methodologies can be applied to any industrial application where the customer's operating specification require a high level of reliability.

In the study, direct tensile test was conducted to perform statistical approaches to the maximum tensile strength of design strengths (120, 150 and 180 MPa), and then the direct tensile database construction was made to examine the reliability of the direct tensile test.

Among agronomists, there appears to be a confusion in selecting among standard deviation (SD), standard error (SE) and confidence interval (CI) in reporting their results as figures and graphs. If there is a confusion in selection among them, there should also be difficulties in interpreting results published in peer-reviewed journals. This review paper aims to help researchers better suited for reporting their results as well as interpreting others by revisiting the definition of SD, SE and CI and explaining in plain words the concepts behind the formula. A variation among observation obtained from an experiment can be explained by the use of SD, a descriptive statistic. If one wants to draw an attention to a variation observed among plant germplasm collected from different regions or countries, SD can be reported along with the mean so that readers can get an idea how much variation exists in the particular set of germplasm. When the purpose of reporting experiment results is about inferring true mean of the population, it is advised to use SE or CI, both inferential statistics. For example, a certain chemical compound is to be quantified from plant materials, estimated mean with SD does not tell the range where the true mean content of the chemical compound would lie. It merely indicates how variable the measured values were from replications. In this case, it would be better to report the mean with SE or CI. The author recommends the use of CI over SE since CI is a sort of adjusted SE. The adjustment comes from t value that considers not only the probability but also n size.