The aim of this study was to develop, digitally record, evaluate, and psychometrically equate a set of Taiwan Mandarin bisyllabic word lists to be used for word recognition testing. Frequently used bisyllabic words were selected and digitally recorded by male and female talkers of Taiwan Mandarin. Twenty normally hearing subjects were presented each word to find the percentage of words which they could correctly recognize. Each word was measured at 10 intensity levels (-5 to 40 dB HL) in increments of 5 dB. Logistic regression was used to include 200 words with the steepest logistic regression slopes in four psychometrically equivalent word lists of 50 words each with eight half-lists of 25 words each. Digital recordings of the psychometrically equivalent bisyllabic word recognition lists are available on compact disc.

The purpose of this study was to create psychometrically equivalent word lists in the language of Cantonese for word recognition testing. Frequently used bisyllabic Cantonese words were recorded by a native female and male talker. The word lists were evaluated by administering the word recognition lists to 20 native speakers of Cantonese with normal hearing. Each list was presented at 10 different intensity levels ranging from -5 to 40 dB HL in 5 dB increments. Logistic regression was used to determine the words with the steepest logistic regression slopes. The 200 words with the steepest slopes were then formulated into four lists of 50 words and eight half-lists of 25 words. The mean psychometric slope value at the 50% location for the male talker was 7.5%/dB while the mean slope for the female talker was slightly steeper at 7.6%/dB. The word lists were digitally recorded on compact discs for worldwide use.

Många företag drabbas regelbundet av churn, ett tillstånd som innebär att existerande kunder slutar handla hos företaget eller använda företagets tjänster för att istället vända sig till konkurrenter. För att säkerställa lojalitet bland kunderna behöver företag därför etablera metoder för att tidigt vinna kundens tillit. Med hjälp av maskininlärning kan processen att identifiera churn automatiseras, så kallad churnprediktion. Mycket forskning finns kring churnprediktion, framförallt inom telekomsektorn och inom företag som erbjuder prenumerationstjänster. Majoriteten av tidigare exempel bygger dock på kunddata som samlats in från flera tidpunkter och syftar till att predicera churn inom en längre tidsperiod, vanligtvis inom ett år. Det finns färre exempel inom kontexten e-handeln, samt forskning om hur maskininlärning kan tillämpas för att enbart utifrån data från kundens första köp och inom en kortare tidsperiod identifiera churn. I denna studie har två maskininlärningsmodeller utvecklats baserat på Random Forest-algoritmen och Logistisk Regression-algoritmen. Syftet var att undersöka vilken algoritm som är bäst lämpad för att predicera om en given kund kommer handla igen eller inte inom en tremånadersperiod, enbart med data från kundens första köp. Undersökningen baserades på data från ett svenskt e-handelsföretag. Modellerna utvärderades med mått för klassificeringsproblem, bland annat Cohen’s kappa och AUC. Trots att Logistisk regression visar sig prestera något bättre tyder resultaten på att båda modellerna har generellt svårt att avgöra om kunden kommer utsätta företaget för churn eller ej. En möjlig förklaring anses vara datamängdens restriktivitet som endast innehåller data från kundens första köp. Däremot konstateras båda modellernas möjlighet att filtrera ut kunder som löper hög risk att utsätta företaget för churn, där Random Forest visar sig vara något bättre på detta. Slutligen konstaterades att modellerna inte påvisar kraftig förbättring jämfört med en naiv lösning där alla kunder antas utsätta företaget för churn, men eftersom även små förbättringar innebär att företaget kan spara pengar kan dock modellernas användbarhet motiveras. / Companies are continuously affected by churn, a condition where existing customers turn to competitors instead using the company’s services. To ensure customer loyalty, it is vital for the company to establish methods to gain the customers trust early on. With the help of machine learning, the process for identifying churn can be automated, known as churn prediction. Research on churn prediction is abundant, especially concerning the telecom sector and subscription-based services. Most of these articles, however, are based on additional, historical data surrounding the customer, aiming to predict churn within a longer time frame, usually a year. The articles focusing on e-commerce, combined with how machine learning can be applied to identify churn within a short period, based solely on data from the customer’s first purchase, are scarce. Two machine learning models are developed based on the Random Forest-algorithm and the Logistic Regression-algorithm. These are tested to see which algorithm is best suited for predicting whether a given customer will buy again or not within a three-month period, with only data from the customer's first purchase from a Swedish e-commerce company. The models were then evaluated with classification metrics, including Cohen’s kappa and AUC. Despite the fact that Logistic Regression performed slightly better, the results showed that both models struggled with the churn prediction. A possible explanation is the restrictiveness of the data set. However, with the option of changing the calibration points on the models’ confidence, allowing the filtration of customers who have a greater chance of leading to churn, both models performed better with Random Forest being slightly superior. The models are considered a slight improvement to a naïve solution where all customers are treated as possible churn. They are also useful given the context, where even minor prevention of churn can lead to profit for the company.

Prenumerationstjänster blir alltmer populära i dagens samhälle. En av nycklarna för att lyckas med en prenumerationsbaserad affärsmodell är att minimera kundbortfall (eng. churn), dvs. kunder som avslutar sin prenumeration inom en viss tidsperiod. I och med den ökande digitaliseringen, är det nu enklare att samla in data än någonsin tidigare. Samtidigt växer maskininlärning snabbt och blir alltmer lättillgängligt, vilket möjliggör nya infallsvinklar på problemlösning. Denna rapport kommer testa och utvärdera ett försök att förutsäga kundbortfall med hjälp av maskininlärning, baserat på kunddata från ett företag med en prenumerationsbaserad affärsmodell där prenumeranten får besöka live-event till en fast månadskostnad. De maskininlärningsmodeller som användes i testerna var Random Forests, Support Vector Machines, Logistic Regression, och Neural Networks som alla tränades med användardata från företaget. Modellerna gav ett slutligt träffsäkerhetsresultat i spannet mellan 73,7 % och 76,7 %. Därutöver tenderade modellerna att ge ett högre resultat för precision och täckning gällande att klassificera kunder som sagt upp sin prenumeration än för de som fortfarande var aktiva. Dessutom kunde det konstateras att de kundegenskaper som hade störst inverkan på klassifikationen var ”Använda Biljetter” och ”Längd på Prenumeration”. Slutligen kommer det i denna rapport diskuteras hur informationen angående vilka kunder som sannolikt kommer avsluta sin prenumeration kan användas ur ett mer affärsmässigt perspektiv. / In today’s world subscription-based online services are becoming increasingly popular. One of the keys to success in a subscription-based business model is to minimize churn, i.e. customer canceling their subscriptions. Due to the digitalization of the world, data is easier to collect than ever before. At the same time machine learning is growing and is made more available. That opens up new possibilities to solve different problems with the use of machine learning. This paper will test and evaluate a machine learning approach to churn prediction, based on the user data from a company with an online subscription service letting the user attend live shows to a fixed price. To perform the tests different machine learning models were used, both individually and combined. The models were Random Forests, Support Vector Machines, Logistic Regression and Neural Networks. In order to train them a data set containing either active or churned users was provided. Eventually the models returned accuracy results ranging from 73.7 % to 76.7 % when classifying churners based on their activity data. Furthermore, the models turned out to have higher scores for precision and recall for classifying the churners than the non-churners. In addition, the features that had the most impact on the model regarding the classification were Tickets Used and Length of Subscription. Moreover, this paper will discuss how churn prediction can be used from a business perspective.

Driving safety has been an issue of great concern in the United States throughout the years. According to the National Center for Statistics and Analysis (NCSA), in 2003 alone, there were 6,267,000 crashes in the U.S. from which 1,915,000 were injury crashes, including 38,764 fatal crashes and 43,220 human casualties. The U.S. Department of Transportation spends millions of dollars every year on research that aims to improve roadway safety and decrease the number of traffic collisions. In spring 2002, the Center for Advanced Traffic System Simulation (CATSS), at the University of Central Florida, acquired a sophisticated reconfigurable driving simulator. This simulator, which consists of a late model truck cab, or passenger vehicle cab, mounted on a motion base capable of operation with six degrees of freedom, is a great tool for traffic studies. Two applications of the simulator are to study the contribution of Light Truck Vehicles (LTVs) to potential rear-end collisions, the most common type of crashes, which account for about a third of the U.S. traffic crashes, and the involvement of Larger Size Vehicles (LSVs) in red light running. LTVs can obstruct horizontal visibility for the following car driver and has been a major issue, especially at unsignalized intersections. The sudden stop of an LTV, in the shadow of the blindness of the succeeding car driver, may deprive the following vehicle of a sufficient response time, leading to high probability of a rear-end collision. As for LSVs, they can obstruct the vertical visibility of the traffic light for the succeeding car driver on signalized intersection producing a potential red light running for the latter. Two sub-scenarios were developed in the UCF driving simulator for each the vertical and horizontal visibility blockage scenarios. The first sub-scenario is the base sub-scenario for both scenarios, where the simulator car follows a passenger car, and the second sub-scenario is the test sub-scenario, where the simulator car follows an LTV for the horizontal visibility blockage scenario and an LSV for the vertical visibility blockage scenario. A suggested solution for the vertical visibility blockage of the traffic light problem that consisted of adding a traffic signal pole on the right side of the road was also designed in the driving simulator. The results showed that LTVs produce more rear-end collisions at unsignalized intersections due to the horizontal visibility blockage and following car drivers' behavior. The results also showed that LSVs contribute significantly to red light running on signalized intersections and that the addition of a traffic signal pole on the right side of the road reduces the red light running probability.

Traffic crashes have been identified as one of the main causes of death in the US, making road safety a high priority issue that needs urgent attention. Recognizing the fact that more and effective research has to be done in this area, this thesis aims mainly at developing different statistical models related to the road safety. The thesis includes three main sections: 1) overall crash frequency analysis using negative binomial models, 2) seemingly unrelated negative binomial (SUNB) models for different categories of crashes divided based on type of crash, or condition in which they occur, 3) safety models to determine the probability of crash occurrence, including a rainfall index that has been estimated using a logistic regression model. The study corridor is a 36.25 mile stretch of Interstate 4 in Central Florida. For the first two sections, crash cases from 1999 through 2002 were considered. Conventionally most of the crash frequency analysis model all crashes, instead of dividing them based on type of crash, peaking conditions, availability of light, severity, or pavement condition, etc. Also researchers traditionally used AADT to represent traffic volumes in their models. These two cases are examples of macroscopic crash frequency modeling. To investigate the microscopic models, and to identify the significant factors related to crash occurrence, a preliminary study (first analysis) explored the use of microscopic traffic volumes related to crash occurrence by comparing AADT/VMT with five to twenty minute volumes immediately preceding the crash. It was found that the volumes just before the time of crash occurrence proved to be a better predictor of crash frequency than AADT. The results also showed that road curvature, median type, number of lanes, pavement surface type and presence of on/off-ramps are among the significant factors that contribute to crash occurrence. In the second analysis various possible crash categories were prepared to exactly identify the factors related to them, using various roadway, geometric, and microscopic traffic variables. Five different categories are prepared based on a common platform, e.g. type of crash. They are: 1) Multiple and Single vehicle crashes, 2) Peak and Off-peak crashes, 3) Dry and Wet pavement crashes, 4) Daytime and Dark hour crashes, and 5) Property Damage Only (PDO) and Injury crashes. Each of the above mentioned models in each category are estimated separately. To account for the correlation between the disturbance terms arising from omitted variables between any two models in a category, seemingly unrelated negative binomial (SUNB) regression was used, and then the models in each category were estimated simultaneously. SUNB estimation proved to be advantageous for two categories: Category 1, and Category 4. Road curvature and presence of On-ramps/Off-ramps were found to be the important factors, which can be related to every crash category. AADT was also found to be significant in all the models except for the single vehicle crash model. Median type and pavement surface type were among the other important factors causing crashes. It can be stated that the group of factors found in the model considering all crashes is a superset of the factors that were found in individual crash categories. The third analysis dealt with the development of a logistic regression model to obtain the weather condition at a given time and location on I-4 in Central Florida so that this information can be used in traffic safety analyses, because of the lack of weather monitoring stations in the study area. To prove the worthiness of the weather information obtained form the analysis, the same weather information was used in a safety model developed by Abdel-Aty et al., 2004. It was also proved that the inclusion of weather information actually improved the safety model with better prediction accuracy.

In recent years the use of advanced driving simulators has increased in the transportation engineering field especially in evaluating safety countermeasures. The driving simulator at UCF is a high fidelity simulator with six degrees of freedom. This research aims at validating the simulator in terms of speed and safety with the intention of using it as a test bed for high risk locations and to use it in developing traffic safety countermeasures. The Simulator replicates a real world signalized intersection (Alafaya trail (SR-434) and Colonial Drive (SR-50)). A total of sixty one subjects of age ranging from sixteen to sixty years were recruited to drive the simulator for the experiment, which consists of eight scenarios. This research validates the driving simulator for speed, safety and visual aspects. Based on the overall comparisons of speed between the simulated results and the real world, it was concluded that the UCF driving simulator is a valid tool for traffic studies related to driving speed behavior. Based on statistical analysis conducted on the experiment results, it is concluded that SR-434 northbound right turn lane and SR-50 eastbound through lanes have a higher rear-end crash risk than that at SR-50 westbound right turn lane and SR-434 northbound through lanes, respectively. This conforms to the risk of rear-end crashes observed at the actual intersection. Therefore, the simulator is validated for using it as an effective tool for traffic safety studies to test high-risk intersection locations. The driving simulator is also validated for physical and visual aspects of the intersection as 87.10% of the subjects recognized the intersection and were of the opinion that the replicated intersection was good enough or realistic. A binary logistic regression model was estimated and was used to quantify the relative rear-end crash risk at through lanes. It was found that in terms of rear-end crash risk SR50 east- bound approach is 23.67% riskier than the SR434 north-bound approach.

This dissertation research aims to develop systemic methods to analyze mental disorder and social media use data in young adults in a dynamic way. The first part of the dissertation is a comprehensive review on modeling methods of longitudinal data. The second part describes the methods that we used to identify symptom clusters that can characterize treatment trajectories and to predict responses of anti-depressants for depression patients. Manhattan distance and bottom-up hierarchical clustering methods were used to identify the symptom clusters. Penalized logistic regressions were conducted to identify top baseline predictors of treatment outcomes. The third part presents of Tweedie distribution application with generalized linear models and growth mixed models for analyzing association between social media use patterns and mental health status. The fourth part is future work and research directions.

It is believed that a large number of people experience remaining symptoms after COVID-19, so-called post-COVID. The formal definition and diagnostic criteria of post-COVID have been a scientific controversy. So far, there is no reliable system for distinguishing the severity of post-COVID. This type of measurement would be helpful in future targeted therapies. Therefore, this thesis aims to evaluate the relationship between an individual’s functional status today and the symptoms present as well as identify relevant groups of post-COVID based on these 17 long-term symptoms of post-COVID. Further, to produce a model for which of these groups an individual belongs to. By using cluster analysis and ordinal logistic regression, Post-COVID Syndrome scores are produced. That is based upon both subjects who were hospitalised and those who were not, collected through a project called COMBAT post-covid. The individuals are then divided into groups based on these scores, and a prediction model is made using ordinal logistic regression and backward deletion. Three well-separated groups of post-COVID are found based on the produced scores. The prediction model indicates that the nine variables Sex, BMI, Smoking, Snuff, Heart disease, Lung disease, Diabetes, Chronic pain and Symptom severity at the onset seem important for predicting someone’s group. This study showed that the remaining symptoms affected an individual’s functional status, including self-reported working ability and general health.

In today’s information era, many news outlets are competing for attention. One way to cut through the noise is to use images. Obtaining images can be both time-consuming and expen- sive for smaller news agencies. In collaboration with the Swedish news agency Newsworthy, we investigate the possibilities of using AI-generated images in a journalistic context. Using images generated with the text-to-image generation model Stable Diffusion, we aim to answer the research question How do the parameters in Stable Diffusion affect the applicability of the generated images for journalistic purposes? A total of 511 images are generated with different Stable Diffusion parameter settings and rated on a scale of 1-5 by three journalists at Newswor- thy. The data is analyzed using ordinal logistic regression. The results suggest that the optimal value for the Stable Diffusion parameter classifier-free guidance is around 10-12, the default 50 iterations are sufficient, and keywords do not significantly affect the image outcome. The parameter that has the single greatest effect on the outcome is the prompt. Thus, to generate photo-realistic images that can be used in a journalistic context, most thought and effort should be put towards formulating a suitable prompt.