To improve Taiwan's competitiveness, Taiwan's government actively encourages businesses to commit to innovative R&D activities by implementing R&D subsidies and incentives. They hope to accumulate intellectual capitals and nurture enough technical professionals to promote industrial upgrades and stimulate economic development. When applying for R&D subsidy, the application must go through the proper approval process. Approved applications will be awarded with substantial financial assistance to fuel further innovations such that R&D subsidy has become critical to many companies. This study used the "Conventional Industry Technology Development"(CITD) as an example and classifies the R&D subsidy applications for analysis based on the 5 variables: company's basic information, program type, commitment to R&D, past experience in applying for R&D subsidy, and strategic alliance with other institutions. Logistic regression is expected to sum up factors that are significant in obtaining R&D subsidies and formulate a predictive model. This enables the government to re-examine its policies and understand the conventional manufacturers' commitment to R&D. Furthermore, the study may assist the companies to assess their chances in obtaining R&D subsidies and serve as a reference for future endeavors. The study suggests that the factors most critical for companies to obtain R&D subsidies are the number of employees, capital size, number of government subsidies already obtained, number of companies non-R&D outsourced to, whether R&D intensity has increased for the past 2 years, and whether R&D funding has increased for the past 2 years. Although plan's duration and non-R&D outsourced dollar amount may have some influence, their impact was not obvious in the model.

The paper constructs recidivism prediction models for women drug prisoners, using the 10 factors evaluated in "drug recidivism risk assessment form" by correctional institutions and the 18 factors studied in the literature. With the new recidivism prediction model, I hope to help improving the prediction accuracy of women drug prisoners¡¦ recidivism. The sample in the paper includes 1,029 drug prisoners released from Kaohsiung Women's Prison between 2008 and 2011. All criminal records are traced until the end of 2011. Two sets of potential risk factors of recidivism are considered in the paper. The first set only contains the factors in the evaluation form, and the second set includes all relevant factors. Using Logistic Regression Analysis and Survival Analysis, the effects of potential risk factors on recidivism are examined. I also predict the probability and the time interval of recidivism. Using the Logistic regression model with the risk factors only in the evaluation form, 58.4% of recidivism can be correctly predicted. While extending the set of potential risk factors, the screening rate of recidivism can be enhanced to 73.3%. The median forecast results are far superior to the average forecast in Survival Analysis. With the potential risk factors in the evaluation form, the difference of predicted recidivism date and the actual date is less than 60 days and less than 180 days in 2.5% and 9.6% of sample respectively. With all relevant risk factors, prediction, the share of sample whose difference of predicted recidivism date and the actual date is less than 60 days and less than 180 days are significantly improved to 10.2% and 27.3% respectively.

The purpose of this research was to create measures and models for the evaluation of Open Source Software (OSS) projects. An exploratory analysis of the development and maintenance processes in OSS was conducted for this purpose. Data mining and text mining techniques were used to discover knowledge from transactional datasets maintained on OSS projects. Large and comprehensive datasets were used to formulate, test and validate the models. A new multidimensional measure of OSS project performance, called project viability was defined and validated. A theoretical and empirical measurement framework was used to evaluate the new measure. OSS project data from SourceForge.net was used to validate the new measure. Results indicated that project viability is a measure of the performance of OSS projects. Three models were then created for each dimension of project viability. Multiple data mining techniques were used to create the models. Variables identified from process, product, resource and end-user characteristics of the project were used. The use of new variables created through text mining improved the performance of the models. The first model was created for OSS projects in the development phase. The results indicated that end-user involvement could play a significant role in the development of OSS projects. It was also discovered that certain types of projects are more suitable for development in OSS communities. The second model was developed for OSS projects in their maintenance phase. A two-stage model for maintenance performance was selected. The results indicated that high project usage and usefulness could improve the maintenance performance of OSS projects. The third model was developed to investigate the affects of maintenance activities on the project internal structure. Maintenance data for Linux project was used to develop a new taxonomy for OSS maintenance patches. These results were then used to study the affects of various types of patches on the internal structure of the software. It was found that performing proactive maintenance on the software moderates its internal structure.

In order to reduce default risk, financial institutions have been investigating into credit ratings of companies, which they want to give credit to. This research tries to give a method for financial institutions to differentiate between default and normal company with financial ratios, which is already announced in their seasonal financial reports. The samples are abstracted from security markets, and restricted to building companies. With Discriminant analysis and Logistic regression models, financial institutions can estimate what company may become into default situation and others stay in good condition. According to this research, financial ratios that can be used to discriminate between default and normal companies are: net worth ratio and short-turn borrowing/liquid asset and asset turnover and gross profit margin. It can also be described with asset turnover and gross profit margin if default risk is been estimated.

The Company often enlarge its scale to maintain its competitive advantage by investing. When company lacks of internal funds, it will raise funds from outside. The purpose of this study is to explore how company chooses financial instruments and influence of the announcement effect on stock price. This study analyzes Taiwan listed company by the the sample period from 1993 to 2005. There are two parts of the thesis. The first is the factor of choosing certain financial instrument. We use logistic regression model, both binary and multinomial, to figure it out. The second is the influence of the announcement effect has on the stock price. We use event study to find whether abnormal return exists. Conclusion: 1. If the company¡¦s size is larger, it will choose debt to raise funds. 2. If R&D expense relative to net sales, debt ratio, the proportion of intangible asset are higher, the company will be tend to raise funds by choosing convertible bond 3. If the stock price is overvalued, the company will choose stock. 4. Taiwan listed company will experience negative stock return whatever it chooses stock, debt, or convertible bond.

This study aims to construct a predictive model that investigates patterning of settlement mounds by employing environmental variables. The results then will help to search for unknown sites of the same age. The methodology is applied to the Lake District of Anatolia for the period of 9000B.C. &amp / #8211 / 5500B.C. Four main sets of data are used in this study. The first set is the settlement data, which includes the names, coordinates, and periods of the sites. The sources of independent datasets are topography, lithology and soil. The study starts with the straightforward procedure of plotting the sites in the region. Then the layers (independent variables), populated with their sub-fields, are included in the model in the GIS to construct a predictive model by using logistic regression. Results reveal some high potential areas with no known occupation, as well as some zones which need more research. Also, hierarchy of environmental variables is detected, which affected the settlement patterning of the study area.

In this thesis, we study improvement of product quality in manufacturing industry by identifying and optimizing influential process variables that cause defects on the items produced. Real data provided by a manufacturing company from the metal casting industry were studied. Two well-known approaches, logistic regression and decision trees, were used to model the relationship between process variables and defect types. The approaches used were compared.

In today&lsquo / s world, knowledge is the most powerful factor for the success of the organizations. One of the most important resources to reach this knowledge is the huge data stored in their databases. In the analysis of this data, DM techniques are essentially used. In this thesis, firstly, a comprehensive literature review on DM techniques for the quality improvement in manufacturing is presented. Then one of these techniques is applied on a case study. In the case study, the customer quality perception data for driver seat quality is analyzed. Decision tree approach is implemented to identify the most influential variables on the satisfaction of customers regarding the comfort of the driver seat. Results obtained are compared to those of logistic regression analysis implemented in another study.

In vehicles one of the most important components which affect comfort of the driver and the purchasing decision is the driver&rsquo / s seat. In order to improve design of a driver seat in a leader company of automotive sector, a comprehensive analysis of customer expectations from the driver seat is performed with a cross functional team formed by representatives of design, marketing, production, quality and services departments. In this study, collection of customer voice data and development of an exceptional &ldquo / customer satisfaction estimation model&rdquo / using these data are presented. Data are modeled by the help of Logistic Regression. This model is able to estimate how much a given customer is likely to be satisfied with the driver seat at a certain confidence level. It is also explained how this model can be used to identify design improvement opportunities that help increase the probability that a customer likes the driver seat. The modeling and analysis approach used for the particular case is applicable in general to many other cases of product improvement or development.

Infant mortality rate is used as an indicator of a nation&rsquo / s economic welfare. Despite the tremendous reduction since 1900s infant mortality rate is still high for developing countries. Infant mortality is reduced from 67 to 21 per 1000 live births in 17 years from 1990 to 2007 in Turkey. However, IMR in Turkey is still much higher than the rates in developing countries which is reported as 5 in 2007. In this thesis, I examine regional, household and individual level characteristics that are associated with infant mortality. For this purpose survival analysis is used in this analysis. The data come from 2003-2004 Turkey Demographic and Health Survey that includes detailed information of 8,075 ever married women between the ages 15-49. 7,360 mothers of these women gave birth to 22,443 children. The results of the logistic regression show that intervals between the births of the infants are associated with infant mortality at lower levels of wealth index. Children from poorer families with preceding birth interval shorter than 14 months or children whose mothers experience a subsequent birth fare badly. Breastfeeding is important for the survival chance of the infants under the age 3 months. Place of delivery and source of water the family uses are also found to be correlated with infant mortality risk. Curvilinear relation between maternal age at birth and infant mortality risk is observed, indicating higher risk for teenage mothers and mothers having children at older ages.