Despite the entry of many media services, television remains the most used media service and accounts for the largest advertising spending globally. One of the main metrics for measuring the successfulness of a television advertising campaign is reach, the percentage of the intended target audience that has seen the television advertisement. To help plan television advertisements, the industry aims to find new methods for predicting television advertisement reach more accurately. Therefore, it is of interest to explore the possibility to utilize machine learning regression models. This report examines how well four machine learning regression models are suited for predicting reach based on historical campaign data. The results indicate that the best-performing model is an XGBoost model with a mean absolute percentage error just below 5%. The report also describes which features impact reach the most and if data augmentation can improve the performance of the machine learning models.

No description available.

No description available.

Early methods of optical fiber communication systems haven't been much promising in terms of efficiency. The presence of various impairments in the fiber channel has forced researchers to uncover solutions in order to minimize those effects. With the advancement of technology, optical solutions were finally easier to implement in the system. To this day, optical compensation methods are still found to be as the best way to minimize fiber impairments. However, such technique does introduce enormous complexity to the system, in addition to a large cost. For that reason, the main focus had to shift to an alternative method. Electrical compensation techniques have provided the factor of simplicity to the optical communication system, not to mention that they are relatively cheaper than optical compensators. Furthermore, electrical schemes were found to handle fiber impairments in a relatively efficient manner. In this thesis, an optical fiber communication scheme using the direct-detection method is simulated. A frequency shifter in the optical domain will be used for the system to have a coherent like detection. At the receiver's side, a linear equalizer is realized to offset the linear effects caused by the fiber. To our knowledge, this will mark the first direct detection transmission system to pass the one thousand kilometre mark in fiber length. Furthermore, we simulate another optical fiber communication design using the coherent detection. A nonlinear compensator adapting the Volterra approach will be used to offset nonlinear impairments. Such performance will be compared to that of a linear compensator. Design trade-offs will be analyzed, and the nonlinear compensator is found to a improve performance when a dispersion compensation fiber (DCF) is introduced in the optical domain. / Thesis / Master of Applied Science (MASc)

BACKGROUND: Understanding the reach and representativeness of participants enrolled in behavioral trials, including nutrition and physical activity trials, helps inform the generalizability of study findings and potential public health impacts. Exploring the reach and representativeness of trials that target low socioeconomic and low health literate participants in rural and medically underserved areas, such as southwest Virginia (SWVA), is especially important. The proposed research is part of Talking Health, a six-month, pragmatic randomized-control trial aimed at decreasing sugar-sweetened beverage (SSB) consumption (SIPsmartER) as compared to matched contact control targeting improving physical activity (MoveMore). This community-based trial targets an 8-county region in SWVA. OBJECTIVES: Guided by the reach dimension of the RE-AIM framework, the primary objectives of this study were to determine if eligible and enrolled participants in the Talking Health trial were representative of: 1) eligible, but declined participants, and 2) the broader targeted 8-county region based on 2010 US county level census data. We hypothesized that eligible and enrolled participants would be represented in terms of age, race, ethnicity, educational attainment, income, and health literacy when compared to eligible and declined participants, as well as to the broader US census data. We also hypothesized that males would be underrepresented. METHODS: Eligibility requirements for the study included being 18 years of age or older, having reliable access to a telephone, drinking 200 kilocalories of SSB per day, and being a resident of SWVA. A variety of recruitment strategies were used such as active recruitment at health departments, free clinics, and local businesses with help from Virginia Cooperative Extension agents along with passive methods such as flyers, newspaper ads, and word of mouth. The eligibility screener included basic demographic information such as gender, age, race, marital status, occupation, income, educational attainment, number of children in household, and insurance provider. The screener also had three validated subjective health literacy questions. Statistical analysis included descriptive statistics, independent sample t-tests, Chi-square tests, and One Way ANOVA tests to examine the representativeness of enrolled participants. RESULTS: In total, 1,056 participants were screened, 620 were eligible (58.7%), and 301 (48.5%) enrolled. On average, demographic data for enrolled participants included: 93% Caucasian; 81.4% female; income of $23,173±$17,144; 32% high school (HS) education; and health literacy score 4.5±2.2(3=High, 15=Low). Among eligible participants, when comparing enrolled vs. declined participants there were significant differences (p<0.05) in educational attainment [enrolled=32% HS, declined=48% HS], health literacy scores [enrolled=4.5(2.2), declined=5.0(3.1)], gender [enrolled=81% female, declined=73% female], age [enrolled=41.8(13.4) years, declined=38.3(13.6) years], and race [enrolled=93% white, declined=88% white]. However there were no significant differences in ethnicity and income. When compared to average US Census data across the eight counties, enrolled participants had a higher educational attainment [enrolled sample=68%HS, Census=58%HS], higher proportion of females [enrolled sample=81%, Census=48%], and lower mean income [enrolled sample= $23,173, Census=$36,675]. There were no meaningful differences in terms of race and ethnicity between the enrolled sample and Census data. DISCUSSION: Contrary to our hypothesis, eligible and enrolled participants differed from non-enrolled participants in terms of age, race, education, and health literacy. Our enrolled sample was slightly older, predominately Caucasian, with higher educational attainment and higher health literacy. However, as hypothesized, there were no significant differences for ethnicity and income status, and men were underrepresented. When the study sample was compared to US Census data, the sample was well represented in terms of age, race, and ethnicity; however, enrolled participants had a much lower average annual income and a higher educational attainment. Men were also underrepresented when compared to the census data. There was no census data to compare health literacy status, which limits information regarding the representativeness of the enrolled sample. Importantly, this study has revealed the representativeness of individuals enrolled in this behavioral trial, helps inform the generalizability of study findings, and identifies future research for community-based studies targeting rural and medically underserved areas in SWVA. For example, future behavioral interventions need concerted recruitment strategies to target males, individuals with lower health literacy status, and individuals with less than a high school degree. Exploring and addressing barriers for study enrollment among these sub-groups is also important. / Master of Science

Because of the prevalence and associated health-care costs of obesity, it is widely recognized today as a serious public health issue. The Affordable Care Act (ACA) passed in 2010 includes provisions for all adults to be screened and behavioral intervention offered to those with a body mass index (BMI) of > 30 kg/m2. While the measure is intended to improve access to weight-loss resources for millions of Americans affected, it may not achieve its intended effect in part because of the inability of disadvantaged populations for whom the burden of obesity is the greatest to access ACA-supported services. The objectives of this study were to identify the characteristics that, paired with an evidence-based weight-loss intervention, will have the highest potential reach while achieving a clinically meaningful weight loss, and whether likely participation differs by gender, race, ethnicity or socioeconomic status. Using characteristics from the most effective adult weight-loss studies that use technology to expand reach, three alternative interventions were examined. Findings show that, while the majority of participants favor the traditional ACA face-to-face model, a program delivered completely online may appeal to racial and ethnic minority groups, which were more likely to choose alternative program structures. Alternative program structures should continue to be explored as an important next step toward developing viable referral options for primary care physicians and could have far-reaching impact toward reducing the disparity of obesity and obesity-related disease among disadvantaged populations. / Master of Science

This study was done for ABB’s work for the phasing out of lead in their soft soldering processes and aims to identify possible lead-free substitutes, due to the RoHs directive and the Reach regulation and the fact that lead is a particularly dangerous chemical. The company’s current solder was defined as SnPb. A literature study has been made where information has been reported regarding lead-free alternatives; SnZn, SnAgCu, SnAg, SnCu, SnBi, SnSb and SnIn. Based on the literature review five lead-free solder were ordered and examined to see if these are possible substitutes for the specific soldering processes. The lead free alternative melting points were measured and the solder technicians ranked the lead-free alternatives in comparison with the reference SnPb to decide which alternative was the most fitting. The literature study and the practical tests show which solder are the best for the company from an economic, social and environmental perspective. The conclusion was that the solder SnCu was the option that fitted the company's soldering processes. To clarify that the solder was a better option than SnPb, the substitution had to be seen from an overall perspective, this could for example, be done by a life cycle analysis.

Dans un contexte réglementaire contraignant (REACh), ce projet a pour but de développer une méthode permettant d'évaluer, le plus amont possible du processus de conception, le risque chimique potentiellement présenté par un équipement aéronautique depuis sa livraison jusqu'à sa fin de vie. Le résultat s'intègre dans l'ensemble des paramètres gérés par les concepteurs, dans un but d'amélioration de la traçabilité des substances dangereuses et de maîtrise du risque chimique présenté par l'équipement. Le développement de la méthode s'appuie tant sur l'analyse des modes d'évaluation classiques de risques chimiques, pour différents périmètres, que sur celle des pratiques de conception aéronautiques. En parallèle, des tests in situ en Bureaux d'Etudes sont réalisés itérativement, permettant de définir des outils d'évaluation et d'aide à la décision adaptés aux besoins identifiés des concepteurs. Les travaux aboutissent à la proposition d'une méthode permettant au concepteur de répondre à la problématique risque chimique à partir des seules données issues du processus de conception, rendant cette nouvelle approche miscible à ses activités quotidiennes. / In a restrictive context (REACh regulation), this project aims at developping a method enabling one to evaluate, as soon as possible in the design process, the potential chemical risk related to any aeronautic equipment' s part during their different life cycle steps (from delivery to end of life). Assesment results is integrated among the numerous data that are by designers, in order to improve hazardous substances traceability and equipment's chemical risk control. The method's development lays on the analysis of different chemical risk assessments modes (different perimeters) and the understanding of aeronautical design practices. In situ tests (design offices) have been run in an iterative way, in order to define assessment ans decision making tools responding to designers identified needs. The work done leads to the proposal of a method which enables the designer to cope with chemical risk problematics, only from the data resulting from design process : this new approach can be considered as miscible in designer's daily activities.

Depuis deux décennies, la recherche militaire se focalise sur l'amélioration des critères de performances des explosifs, tout en prenant en compte leurs impacts environnementaux et toxicologiques. Ces enjeux sont encadrés par une réglementation stricte : REACh (Registration, Evaluation, Authorization and Restriction of Chemicals) permettant d'assurer un haut niveau de protection sanitaire et environnementale. De nos jours, développer des explosifs ou molécules hautement énergétiques (High Energy Materials (HEM)) ayant un effet réduit sur l'homme et l'environnement est un sujet de préoccupation majeur. Ainsi, en collaboration avec Airbus Safran Lauchers (ASL), un programme de recherche a été mis en place, afin d'obtenir des outils optimisés pour la prédiction de la toxicité des HEMs et concevoir de nouvelles molécules HEMS non toxiques et réglementaires.Différentes méthodes in silico ont été utilisées dont des Relations Structure Activité Quantitatives (ou Quantitative Structure-Activity Relationship (QSAR)) et le Machine Learning. La recherche de similarité structurale parmi les molécules est un outil novateur sur lequel nous avons basé nos prédictions in silico. Cette similarité est obtenue grâce à un algorithme intelligent développé au sein du Pôle Rhône Alpin de Bio-Informatique de Lyon et qui a donné lieu à un brevet. Cet algorithme nous permet d'obtenir des prédictions plus précises basées sur des données expérimentales issues de directives européennes / For the last two decades, the military research has focused on the improvement of explosive performances, while taking into account their environmental and toxicological impacts. These issues are governed by strict regulations: REACh (Registration, Evaluation, Authorization and Restriction of Chemicals) to ensure a high level of health and environmental protection.Today, it's a major consideration to develop High Energetic Materials (HEM) or molecules who's hazard on human health and environment are reduced. Thus, in collaboration with Airbus Safran Lauchers (ASL), a research program was set up to obtain optimized tools for predicting the potential toxicity of HEM and to design new non-toxic and regulatory molecules.Different in silico methods have been used, including Quantitative Structure Activity Activity Relationships (QSARs) and Machine Learning.The search for structural similarity among molecules is an innovative tool on which we based our predictions in silico. This similarity is obtained thanks to an intelligent algorithm developed within the Pole Rhone Alpin de Bio-Informatique of Lyon which gave rise to a patent. This algorithm allows us to obtain more accurate predictions based on experimental data from European directives

Uncertainty about how chemicals affect human health and the environment, and shortcomings in current chemicals legislation, make it necessary to motivate chemical users to work actively with risk management in order to achieve a high level of protection. In this study, the drivers of chemical risk management have been investigated in order to find out how a pro-active work could be promoted. A Q-methodology study was performed online with 15 respondents at 10 different engineering companies in the Gnosjö region in Sweden. The respondents were part of the middle management and the local top management and had an environmental responsibility. Two groups with different views on drivers were identified. To comply with legislation was the main driver of the market-oriented group, while the plant-oriented group considered an improved work environment for employees as the main driver. A tendency to differences between how drivers were valued by middle management and local top management could be distinguished. The study indicated a need to motivate companies to stay ahead of legislation, as well as to enhance the understanding of the environmental impact that the operations have on the environment, and may be used as a basis for further studies of drivers of chemical risk management. / Osäkerhet om hur kemikalier påverkar människors hälsa och miljön samt brister i nuvarande kemikalielagstiftning gör att företag som hanterar kemikalier behöver motiveras till att arbeta aktivt med riskhantering för att en hög skyddsninvå ska kunna uppnås. I föreliggande studie har drivkrafterna bakom arbetet med riskantering av kemikalier undersökts i syfte att ta reda på hur ett proaktivt arbete skulle kunna främjas. En Q-metodologistudie genomfördes online med 15 respondenter på 10 olika verkstadsföretag i Gnosjöregionen i Sverige. Respondenterna utgjordes av personer på mellanchefs- och ledningsgruppsnivå med miljöansvar. Två grupper med olika syn på drivkrafter identifierades. Att uppfylla lagstifning var den viktigaste drivkraften för den marknadsorienterade gruppen, medan den fabriksorienterade gruppen betraktade en förbättrad arbetsmiljö för anställda som den viktigaste drivkraften. En tendens till skillnad i hur drivkrafter värderades på mellanchefs- och ledningsgruppsnivå kunde urskiljas. Studien indikerade ett behov av att motivera företagen till att ligga steget före lagstiftning, samt att öka förståelsen för verksamheternas påverkan på miljön och kan ligga till grund för fortsatta studier rörande drivkrafter bakom riskhantering av kemikalier.