http://ir.tum.ac.ke/handle/123456789/17356 Contains electronic theses & dissertations for this institute 2026-06-13T18:42:35Z http://ir.tum.ac.ke/handle/123456789/17629 ROUTE OPTIMIZATION MODEL FOR SOLID WASTE COLLECTION USING MACHINE LEARNING AND GIS TECHNIQUES JEFWA, TITUS The County Government of Mombasa (CGM) faces a major challenge in solid waste management (SWM), which involves collecting and removing waste. The study aimed to improve the waste transportation and logistics operations from collection points to final disposal by computing an algorithm for the optimal waste collection route in Mombasa. The ultimate goal of this study is to develop a model that optimizes the waste collection system using machine learning and geographical information system (GIS) technologies. The study aimed to determine the optimal waste collection route using ant colony, integer linear programming, and simulated annealing algorithms. These algorithms were used to analyze the data collected on the waste collection process, which included locations of transfer stations, time taken, distance, and cost for fuel. The data was first cleaned and tested for distribution, and the optimal factors were then identified. The supervised machine learning process was then applied to identify the relationships between the variables, make predictions, and find the optimization route. Finally, the integration of geographical information system (GIS) process was used to determine the final optimal route for the waste collection process. The results showed that the use of ant colony, integer linear programming, and simulated annealing algorithms was effective in finding the optimal waste collection route, resulting in a more efficient, cost-effective, and sustainable waste management system. This study is crucial and greatly benefit the department responsible for SWM in the CGM by improving the efficiency and effectiveness of waste collection. It primarily use quantitative research methods and intended to evaluate the current waste collection network. The study developed a map of the existing waste collection network using quantum geographical information system (QGIS) software and use a supervised machine learning model to find the best solution to the route optimization problem, determining the shortest path with the minimum cost 2023-01-01T00:00:00Z http://ir.tum.ac.ke/handle/123456789/17628 A DEEP LEARNING MODEL FOR MICROPLASTICS DETECTION IN OPEN SEWER SYSTEMS MICHAEL, JOSEPH Microplastics (MPs) are small plastic particles that pose a threat to aquatic organisms and human health. Detecting MPs in bodies of water is critical for controlling their flow and limiting their negative effects. This study proposes a Deep Learning algorithm for detecting MPs in photos taken from open sewer systems that flow into the ocean. The research adopted the Sample, Explore, Modify, Model, and Assess (SEMMA) framework, a comprehensive data mining process. A dataset comprising 1000 photos was constructed from locations in Kilifi, Mombasa, and Kwale counties in Kenya, by employing the Scale-Invariant Feature Transform (SIFT) algorithm for feature extraction. The researchers compared the performance of two object detection models: Sing-Shot Detector (SSD), and Convolutional Neural Network (CNN), and discovered that SSD performed best with a mean average precision (mAP) score of 100%, while CNN performed worst with 96.5%. A model for detecting MPs in photos taken from open sewer systems that flow into the Indian Ocean along Kenya's coast was developed using the best-performing SSD model. The model can be used to detect MPs in other open sewer systems, assisting in the implementation of effective management and control measures. Future research could look into creating a mobile app that captures images and provides information about MPs in open sewer systems 2023-01-01T00:00:00Z http://ir.tum.ac.ke/handle/123456789/17627 MACHINE LEARNING MODEL FOR PREDICTION OF POSTPARTUM DEPRESSION, A CASE OF MOMBASA COUNTY MONGARE, GEORGE Postpartum depression is a medical condition which affects many mothers. The condition exposes the mother and the newborn baby to illnesses that can lead to death. Management of the condition requires heavy expenditure incurred by the family, government, and stakeholders. The condition is also a source of many social problems. Manual systems are currently used to predict the condition which is slow and inconsistent. Machine learning technology which has reliably been used in prediction modelling in other domains can also be employed to build a model to predict mothers at risk of postpartum depression during pregnancy for primary prevention. In this study, perinatal records were collected from 324 mothers attending postnatal healthcare clinics at the Coast General Teaching and Referral Hospital and Tudor Sub-County Hospital in Mombasa County. The filter feature in WEKA was used to split the data into 70% and 30% for model training and testing respectively. Models were built on WEKA machine learning platform using logistic regression, support vector machine, extremely randomized trees, random forest and adaptive boosting algorithms which were identified from literature. A positive case of postpartum depression was defined as diagnosis or treatment of major depression within one year after delivery. Random forest model produced the best performance with a receiver operating characteristic (ROC) curve area of 0.863867 which is comparable within the bracket of high performing models. With this level of performance, the model can be used by healthcare staff to make quick and consistent prediction for early mitigation measures. Further research could be done with more data collected from other counties in Kenya 2023-01-01T00:00:00Z http://ir.tum.ac.ke/handle/123456789/17626 A BLOCKCHAIN-BASED MODEL FOR CURBING INSTITUTIONAL ACADEMIC CERTIFICATE FRAUD KABIBI, ESTHER There is need for a certificate authentication mechanism in Africa to solve internal academic certificate fraud activities. This is because some fake certificates in circulation appear to be genuine but were obtained illegally from credible learning institutions. This is caused by an assumption that certificates being issued at the university are authentic. While authentication of the final academic certificate was studied in previous research, many researchers focused on securing the final academic certificate and assumed that all certificates issued by institutions were genuine thus creating a loophole for institutional academic certificate fraud. The purpose of this research was to develop a blockchain based model that curbs institutional academic certificate fraud. The research target population was public universities in Kenya. Sampling was done using simple random sampling to identify universities to participate in the study. Secondary data was used to validate the model while primary data was used to construct the data mapping structure. Primary data was collected from the registry department and examination using questionnaires while secondary data was collected using document review. Smart contracts were written using the GOLANG and deployed to the hypledger fabric. APIs are used to interact with the model to insert or retrieve data. The model has controls (smart contracts) to ensure a student goes through the entire learning process before he is awarded an academic certificate. To validate the model, experiments were carried using the secondary data. The results of the experiments shows that the model prevents internal fraud by making sure that only students registered in the blockchain and who completed the academic requirements can receive their academic certificates. Verification of the academic certificate is done by simply scanning a QR Code embedded on the academic certificate using a developed mobile app. The model only validates certificates whose certificate information was generated by the model. Ensuring that certificates originating from universities are authentic will retain or increase the reputation and credibility of the institution. On the other hand, employers and other interested parties will also give jobs to qualified people thus increasing the throughput and general job performance 2023-01-01T00:00:00Z