Solar energy coupled with energy efficiency awareness can be an important step to achieving thefinal goal of having a sustainable environment in the country. Together with a good energy policyincluding regulations, this goal can be achieved as it will help consumers make more rationaldecisions when it comes to new technology. A key problem identified in this work is the lack ofawareness and strict energy policies. The current funds put in place are not mobile and have not got to a large number of people. Thegovernment’s regulations on the other hand are not strict. This report aims to improve the currentstatus by suggesting strategies to reach out to as many people as possible and educate them aboutsolar energy options in Kenya. It aims to prepare to build a platform that creates awareness onsolar energy, energy efficiency as well as the policies and finances needed to support their uptake. In order to build this platform, analysis has been done through research and meetings withstakeholders who have provided first-hand information in order to formulate the foundation to afuture platform. The information gathered has then been used to explain and meet the challengesfaced in solar energy in Kenya. The conclusion of this project was that majority of the people inKenya lack required knowledge about solar energy. There is need to create awareness in solarenergy.

Artificial Intelligence (AI) has its roots in every field in present scenario. Healthcare is one of the sectors where AI is reaching considerable growth in recent years. Tremendous increase in healthcare data availability and considerable growth in big data analytic methods has paved way for success of AI in healthcare and research is being driven towards improvement in quality of service. Healthcare data is stored in the form of Electronic Health Records (EHR) which consists of temporally ordered patient information. There are many challenges with EHR data like heterogeneity, missing values, biases, noise, temporality etc. This master thesis focuses on addressing the problem of visit level irregularity which refers to irregular timing between events (patient’s visits).

Large microservice clusters deployed in the cloud can be very difficult to both monitor and debug. Monitoring theses clusters is a fi€rst step towards detection of anomalies, deviations from normal behaviour. Anomalies are oft‰en indicators that a component is failing or is about to fail and should hence be detected as soon as possible. Th‘ere are oft‰en lots of metrics available to view. Furthermore, any errors that occur oft‰en propagate to other microservices making it hard to manually locate the root cause of an anomaly, because of this automatic methods are needed to detect and correct the problems. Th‘e goal of this thesis is to create a solution that can automatically monitor a microservice cluster, detect anomalies, and fi€nd a root cause. Th‘e anomaly detection is based on an unsupervised clustering algorithm that learns the normal behaviour of each service and then look for data that falls outside that behaviour. Once an anomaly is detected the proposed method tries to match the data against prede€fined root causes. ‘The proposed solution is evaluated in a real microservice cluster deployed in the cloud, using Kubernetes together with a service mesh and several other tools to help gather metrics and trace requests in the system.

I det här arbetet vill jag utgå från mig själv som en fri hantverkare, arbetare närmredet objekt jag skapar och utforska skärande i syfte att skapa dekorativa element i ett möbelsammanhang. Jag har låtit mig inspireras av historiska filosofer samt hantverkareoch konstnärer som värnar om hantverket som sådant men också för denmänskliga kreativa lusten och glädjen. Min slutsats i arbetet är att jag som hantverkarehar större kontroll och glädje av att arbeta i ett självförverkligande syfte där jagoch mitt görande står i fokus, snarare än slutresultatet av görandet / In this work I want to start from myself as a free craftsman, in my work get closerto the object I create and explore cutting in order to create decorative elements ina furniture context. I have been inspired by historical philosophers as well as craftsmenand artists who care for the craft as such but also for the human creative desireand joy. My conclusion in the work is that, as a craftsman, I have greater controland joy of working for a self-realizing purpose where I and my work are in focus,rather than the end result of doing so.

Plast i naturliga miljöer är ett stort problem i dagens läge då materialet oftast inte är 100 %biologiskt nedbrytbart. Plasten bryts ner till bitar i mikrostorlek, dvs. under 5 mm. Mikroplasten ackumuleras i landmassor likväl som i vattendrag och hav. Eftersom materialet kan bära mycket högre koncentrationer av miljögifter än den naturliga miljön så ökar den halter cancerframkallande- och hormonstörande ämnen hos djur. Plast är bevisligen farligt förbiologisk mångfald och måste minskas.Bioplast är en plast med biologiska komponenter men behöver därför inte vara biologiskt nedbrytbar i den naturliga miljön, som t.ex. PLA, polylaktid. Den plasten är baserad på majs men endast nedbrytar i stränga industriella processer. En annan familj av polymerer som kanskapa 100 % biologiskt nedbrytbar plast är PHA, Polyhydroxyalkanoater. Det innebär att PHA bryts ned till CO2 och vatten inom de sex områdena som omfattar all naturlig miljö. I denna rapport har det undersökts producering av PHA från skogsindustriellt bioslamm. Bioslammet som det har utförst fyra försök på i denna rapport är från reningsverket som hör till Gruvöns pappersbruk. Bioslammet som har tagits är returslam efter sedimentering. Försöken som har utförts har haft temperaturbegränsningar, 20oC och 35oC, för att se hur temperaturen påverkar PHA-ackumuleringen. Det gjordes fyra försök på 10 timmar vardera, dvs. två försök på varje temperatur. Försöken utfördes i cylindriska tankar på 50-, 45-, 31-liter, anledningen att volymen skiljde mellan försöken är för att möjligheten inte alltid fanns att hämta samma mängd bioslam från Gruvön vid varje tillfälle. Försöken utfördes över tre dagar med förberedelser dag 1, utförande dag 2 och FTIR- och TOC-analyser dag 3. Under dag 2 har tillsatser av substrat varit en viktig faktor då det har varit pulsmatning med tidsintervall mellan 30- och 60 minuter. Substratet har varit ättiksyra och natriumacetat då bakterierna använder acetatet som en lättnedbrytbar kolkälla. Skillnaden mellan ättiksyra och natriumacetat är att den senare har en minimal påverkan på pH-värdet vid tillsättning. Aspekter som har undersökts i samband med försöken har varit b.la. syrehalt, pH-värde, temperatur och acetatförbrukning. Syftet har varit att få en insikt i hur temperaturen påverkar bakterierna från en blandad kultur och deras förmåga att ackumulera polyhydroxyalkanoat (PHA). Rapporten ska sedan kunna användas som beslutsunderlag vid framtida PHA-producering samt vidarestudier i ämnet. Målet med studien är att undersöka hur ackumulering av PHA och acetatförbrukningen påverkas vid 20oC respektive 35oC. Resultaten visar att de önskade temperaturerna inte riktigt uppnåddes då de blev 22oC och 33oC. Det ses tydligt att bakterierna förbrukar betydligt mindre mängder substrat vid 22oC än vid 33oC utan att PHA-ackumuleringen påverkas nämnvärt. Skillnaden i ackumuleringshastigheten mellan de två temperaturerna var såpass liten att den anses försumbar. Den maximala PHA-halten uppnåddes tyvärr inte på grund av försökens tidsbegränsning på 10 timmar. Slutsatsen är att lägre temperaturen leder till mindre acetatförbrukning men lika hög PHA-ackumulering, att ackumuleringshastigheten är försumbar samt att 10 timmar ackumuleringstid inte var tillräckligt för att nå maximala halter av PHA. Framtida studier i ämnet borde kolla på följande problemställningar: • SÄ-koncentrationens inverkan på acetatförbrukning och PHA-ackumulering. • Stegrande eller kontinuerlig tillsättning av substrat. • Temperatur och tillsättning av substrat, dess inverkan på syrehalten. • Tidsaspekt vid PHA-ackumulering. Dessa problemställningar har tagits fram utifrån denna rapport, dvs. vilka faktorer som uppmärksammades till att ha potential till att vara viktiga inför PHA-produktion i pilotskala. / Plastic in natural environments is a big problem in today's situation as the material is usuallynot 100 % biodegradable. The plastic is broken down into pieces of micro-size, meaning below 5 mm. The micro-plastic accumulates in landmasses as well as in watercourses and oceans. Because the material can carry much higher concentrations of environmental toxins than the natural environment, it increases the levels of carcinogenic and endocrine disruptors in animals. Plastic is apparently dangerous for biodiversity and must be reduced. Bioplastic is a plastic with biological components but therefore does not have to bebiodegradable in the natural environment, such as PLA, polylactide. The plastic is based on corn but only degrades in severe industrial processes. Another family of polymers capable of producing 100 % biodegradable plastic is PHA, Polyhydroxyalkanoates. This means that PHA is broken down into CO2 and water in the six areas that cover all natural environments. In this report, the production of PHA from forest industrial biosludge has been investigated. The biological sludge that has been used in four attempts in this report is from the sewagetreatment plant that belongs to Gruvön's paper mill. The sludge that has been taken is the returnsludge after sedimentation. The attempts that have been performed have had temperature limitations of 20°C and 35°C, to see how the temperature affects PHA accumulation. Four attempts were done in 10 hours each, meaning two attempts at each temperature. The attempts were carried out in cylindrical tanks of 50-, 45-, 31-liter, the reason that the volume differed between the attempts is because the possibility was not always available to retrieve the same amount of bio-sludge from Gruvön on each occasion. The attempts were carried out over three days with preparations on day 1, execution on day 2 and FTIR- and TOC-analyzes on day 3. During day 2, additions of substrate have been an important factor since there has been pulse feeding with time intervals between 30- and 60 minutes. The substrate has been acetic acid and sodium acetate since the bacteria use the acetate as an easily degradable carbon source. The difference between acetic acid and sodium acetate is that the latter has a minimal effect on the pH-value when added. Aspects that have been investigated in connection with the attempts have been e.g. oxygen content, pH-value, temperature and acetate consumption. The purpose has been to gain an insight into how the temperature affects the bacteria from a mixed culture and their ability to accumulate polyhydroxyalkanoate (PHA). The report should then be used as a basis for decision-making in future PHA production and further studies on the subject. The aim of the study is to investigate how the accumulation of PHA and acetate consumptionis affected at 20oC and 35oC, respectively.The results show that the desired temperatures were not really achieved when they became 22°C and 33°C. It is clearly seen that the bacteria consume considerably smaller amounts of substrate at 22oC than at 33oC without the PHA accumulation being significantly affected. The difference in the rate of accumulation between the two temperatures was so small that it is considered negligible. The maximum PHA content was unfortunately not achieved due to the 10 hour attempt time limit. The conclusion is that the lower temperature leads to less acetate consumption but equallyhigh PHA accumulation and that the difference in accumulation rate between the two temperatures is negligible and last that 10 hours of accumulation time was not sufficient to reach maximum levels of PHA. Future studies on the subject should look at the following issues: • The impact of SE concentration on acetate consumption and PHA accumulation. • Increasing or continuous addition of substrate. • Temperature and addition of substrate, its impact on oxygen content. • Time aspect of PHA accumulation. These issues have been developed on the basis of this report, meaning which factors were highlighted to have the potential to be important for PHA production on a pilot scale.

Genetic algorithms (GAs) are a powerful tool for solving multi-objective optimization problems. Resource allocation and scaling of cloud systems typically involve multiple conflicting objectives, such as high through putin the presence of failures, cost, and reduced latency. Microservice-based architectures introduce additional complexities since the underlying services respond differently to different workloads. In this work, the performance of two multi-objective GAs is compared on the problem of finding efficient scaling configurations of a microservice-based architecture. Results show that while the use of GAs is effective at finding efficient configurations, GAs can not be used for larger systems involving many microservices or for systems that make use of caching.

New databases and query-languages are created every year. Developers using these technologies have to learn all the different ways to access the databases. When using a query-language to retrieve information it is easy for the query to become really long and complex.Different ways to implement an abstraction layer on top of databases was investigated. By looking at what is easiest to learn and use a Java implementation was used. In the resulting implementation the developers could create a query-object which is sent into a converter which creates the correct query-syntax. After showing the implementation to different developers at Cinnober they felt that an abstraction layer like this one could be really useful. The solution would provide a uniform way of creating queries and that it would be easier to use and understand once you got used to it. The performance of the implementation was high enough for the standards set by Cinnober, it was also deemed easy to understand which was an important criteria for usage in production. A new way of communicating to databases is presented and could be used in production if developers chooses to improve the current implementation.

Manually supervising a million documents yearly becomes an exhaustive task. A step towards automating this process is to interpret and classify the contents of a document. This thesis examines and evaluates different object detection models in the task to localize and classify multiple objects within a document to find the best model for the situation. The theory of artificial neural networks is explained, and the convolutional neural network and how it is used to perform object detection. Googles Tensorflow Object Detection API is implemented and models and backbones are configured for training and evaluation. Data is collected to construct the data set, and useful metrics to use in the evaluation pipeline are explained. The models are compared both category wise and summarized, using mean average precision (mAP) over several intersection over union (IoU) thresholds, inference time, memory usage, optimal localization recall precision (oLRP) error and using optimized thresholds based on localization, recall and precision. Finally we determine if some model is better suited for certain situations and tasks. When using optimal confidence thresholds different models performed best on different categories. The overall best detector for the task was considered R-FCN inceptionv2 based on its detection performance, speed and memory usage. A reflection of the results and evaluation methods are discussed and strategies for improvement mentioned as future work.

In a Real Time Clearing System (RTCS) there are several thousands of transactions per second, and even more messages are sent back and forth. Th‘e high volume of messages and transactions being sent within the system eventually leads to some anomalies arising. Th‘is thesis examines how to detect such anomalies with unsupervised Machine Learning models such as, Support Vector Machine(SVM) One Class (OC), Isolation Forest (iForest) and Local Outlier Factor(LOF). Th‘e main objective is to investigate if anomaly detection is useable in Cinnobers RTCS, only using unsupervised models and if they perform at an acceptable level. Th‘e evaluation of the models will be done using a rough labeling method to score them on detection rate, F-score and Ma‹hews correlation coecient (MCC). Th‘e results of the thesis shows that SVM OC is the best model of the three, but requires hyper parameter tuning to perform at an acceptab lelevel so that it may be used for the RTCS without human supervision.

The increasing demand from customers and users on systems are constantly increasing, demanding faster more robust solutions. There are no exception especially when it comes to clearing technology in the financial industry, where the system needs to handle information fast and be responsive. The thesis have investigated the potential performance improvements by using Content Delivery Network (CDN) based technology on Cinnobers Real Time Clearing system RTC. The thesis have based on constructing a functional CDN node, and look at the impact that it has on the system in regards of handling reference data. Could this approach help improve the system and improve the performance of the system and scalability in regards to reference data handling. Based on the results that was gathered by comparing the node implementation with the original system, there are clear indications performance improvements on the system. Where the results concluded that with the CDN node implementation a performance improvement when requesting reference data, had an increase in fetch time by a factor of 6. By using the results gathered in the thesis, a simulation was created to simulate the effect of a fully scaled CDN. The simulation concluded that the implementation could reduce latency by 44 minutes during a day of use.