Academic submissions (CoCIS)

Permanent URI for this collection

http://10.29.1.53:4000/handle/10570/4083

Browse

Browsing Academic submissions (CoCIS) by Title

  • Item
    An agricultural knowledge-sharing framework for smallholder farmers and agricultural knowledge experts
    (Makerere University, 2018-04) Mwesigwa, Ezra
    Using Participatory Action Research (PAR) as our research methodology, we selected smallholder farming communities in the districts of Apac, Lira, Kumi and Bukedea with a major goal of developing an ICT based knowledge sharing framework between smallholder farmers and agricultural knowledge experts. In order to better understand and sufficiently study the engagement between smallholder farmers and agricultural knowledge experts, we built a mobile tool that was used as a medium for exchange of knowledge between the farmers and the knowledge experts. A combination of data collected from the field using questionnaires, narrative interviews and mobile tool meta data helped us design the knowledge sharing framework which was later implemented and validated for knowledge sharing. The final framework was achieved by extending a knowledge management framework developed by Heisig (2009). Through our developed framework, we emphasize that successful knowledge sharing is achieved by critically harnessing five (5) enablers to knowledge sharing. Chapter 1 of this thesis presents a comprehensive background on the growing information needs of smallholder farmers worldwide and in Uganda. It also presents justification for this study, the objectives and research questions that guided the researchers. Chapter 2 provides a grounded evidence through literature review about previous scholarly works to address the growing information needs of smallholder farmers. Within this chapter, we review a number of knowledge sharing models and frameworks and compare them to one another. We identified existing gaps in the literature reviewed at the end of the chapter. In Chapter 3, we show in detail how Participatory Action Research (PAR) was used to achieve the study objectives. We divided the study in four major stages i.e. Problem Diagnosis, Solution Design, Solution Implementation and finally Evaluation of the implemented solution. Chapter 4 covers the data analysis process. We used questionnaires and narrative interviews to collect data from which we drew meanings and conclusions. Lastly, Chapter 5 presents discussions, recommendations and conclusion of the study.
  • Item
    Application of random forest regressor algorithm to predict PM2.5 concentration levels in Kampala
    ( 2018-12-06) Wabinyai, Fidel Raja
    As it happens in every society, it is every body's wish to live in a clean and fresh environment. However, this might not be achieved in every daylife but at least the level of pollution can be controlled. Air pollution is one of the leading global public health risks but its magnitude in many developing countries is not known. As is in many African cities, fine particulate matters (PM2.5) is dangerously high in Kampala. This thesis uses data mining algorithms to build a predictive model for the following days PM2.5 concentration level. The prediction of concentrations of pollutants can be a powerful tool in order to take preventive measures such as the reduction of emissions and alerting the affected population. This thesis presents a forecasting model to predict the daily average concentrationof PM2.5 for the next few days(i.e. 3 to 5days). The proposed model used in this thesis was Random Forests regression. Random Forests regressor was compared with 4 other regression models namely Extra Trees Regressor, Gaussian Process Regressor, XGBoost, and Elasticnet. The performance estimation is determined using the Root Mean Square Error (RMSE), the Mean Absolute Error (MAE) and R-squared (R2). The results demonstrated that the Random Forests regressor algorithm outperformed other models. 6 pollution monitoring stations in Kampala measuring PM2.5 were selected. We found that the mean concentration of PM2.5 pollution was 3 times higher than the World Health Organization (WHO) recommended level.
  • Item
    Assessment of records management practices at All Saints Cathedral, Kampala
    (Makerere University, 2022-03-28) Kiconco, Patience Pretty
    A study was carried out to assess records management practices at All Saints Cathedral, Kampala. The objectives of the study were; To establish the existing church records at ASCK; to find out how church records are managed at ASCK; to identify the challenges of church records management at ASCK; to make suggestions for the best church records management practices at ASCK. The study purposefully selected (5) clergy, (11) Heads of Departments, (2) administrators and (2) secretaries as respondents during data collection. The total sample in this study was 20 respondents. Data was collected using non-participatory observation, interviews and documentary review. The study revealed that the types of records at ASCK included; Sacramental records (Marriage records, Baptism records, and Confirmation records), Cathedral Registers, Legal records, Financial records, Human Resource records, Building and Construction records, Cathedral Annual Reports, Announcement Books, Worship and Arts, and artefacts like liturgical vestments, souvenirs and trophies. At ASCK, each office or department kept its own records and were in custody of staff that occupy those offices. Most of the records were seen squeezed and stuffed in boxes, folders and shelves. Others were on top of tables and old cupboards. The study established that was no Records officer at ASCK, there was no records officer nor a Records Registry. There were no policies in place governing records management and there were no resources allocated to records management on the Cathedral Annual Budget. There was also inadequate and insufficient records storage space at ASCK. This study made recommendations to ASCK Management and Staff addressing the aforementioned challenges which included; Recruitment of a Records officer and training staff in Records Management, establishment of a Cathedral Registry or Archive, developing policies and procedures to govern the Records management program, ensuring there is an offsite storage facility, backing up all records, developing a disaster preparedness plan, and developing a retention & Disposal Manual
  • Item
    Augmented continuous wavelet transform features for deep learning-based indoor localization using WiFi RSSI Data
    (Makerere University, 2021-08) Ssekidde, Paul
    Localization in indoor environments is currently one of the challenges in navigation research. The conventional global positioning system (GPS) is affected by weak signal strengths due to high levels of signal interference and fading in indoor environments. Therefore, new positioning solutions tailored for indoor environments need to be developed. In this paper, we propose a deep learning approach for indoor localization. However, the performance of a deep learning system depends on the quality of the feature representation. This paper introduces two novel feature set extractions based on the continuous wavelet transforms (CWT) of the received signal strength indicators' (RSSI) data. The two novel CWT feature sets contain augmented data generated from the CWT of RSSI data with additive white Gaussian noise. The first feature set is an image feature set while the second feature set is a numerical feature set composed of the power spectral densities (PSD) of the CWT that is dimensionally equalized using the principal component analysis (PCA). These proposed images and numerical data feature set were both evaluated using the CNN and ANN model respectively with the goal of identifying the room that the human subject is in and estimating the precision location of a human subject in an indoor environment. Extensive experiments were conducted to generate the proposed augmented CWT feature set and numerical CWT PSD feature set using two analyzing functions namely the Morlet and Morse. The feature sets were both tested to predict a room and the precise position a human subject is at. For validation purposes, the two proposed feature sets were validated against each other and other existing feature set formulations. The accuracy, precision and recall results show that the proposed feature sets perform better than conventional feature sets used to validate the study. Similarly, the mean displacement error generated by the proposed feature set predictions is less than that of the conventional feature sets used in indoor localization. More particularly, the proposed augmented CWT-image feature set out performs the augmented CWT-PSD numerical feature set. The results also show that the Morse-based feature sets trained with CNN produce the best indoor positioning results compared to all Morlet and ANN-based feature set formulations.
  • Item
    Automated intestinal parasite detection in stool samples using custom convolutional neural networks
    (Makerere University, 2022-03-10) Rwakazooba, Ezra Aliija
    Intestinal parasitic infections can cause serious health problems with relatively high infections in the developing world. Microscopy of stool remains the gold standard method for the diagnosis of intestinal parasites. However, this method can be time-consuming, and it is also challenging to maintain consistency in diagnosis across different technicians. This is also hindered by the few competent and skilled technicians in the developing countries where the prevalence of intestinal parasites is high. Deep learning has increasingly gained application ground in different challenging computer vision tasks. There is also growing literature of the use of the same technologies in health diagnostic fields such as microscopy. What is used in the state-of-art computer vision challenges, oftentimes gets applied to real-world challenges. However, this has met different limitations in sensitivity and specificity given the broader range of diversity in data sets; for example, in this study of intestinal parasite detection. In general, deep learning continues to provide good performance to computer vision problems across multiple disciplines. In this work, the use of AlexNet and GoogleNet models’ performance on the diagnosis of intestinal parasite eggs in stool samples is evaluated. This work goes ahead to compare these out-of-the-box fine-tuned models with a custom-trained Convolutional Neural Network on the same task. In all cases, accuracy from the out-of-the-box models is very high with GoogleNet ROC AUC of 0.99 and AlexNet ROC AUC of 1.00, and runs on a very low computing resource system, which speaks to the fact that out-of-box models can re-purposed for real-world health diagnostic challenges.
  • Item
    Comparative evaluation of na ̈ive–bayes and k–nearest neighbor classifiers to improve prediction of short-term precipitation
    (Makerere University, 2021-02-10) Balikuddembe, Joseph Kiwuuwa
    This research carried out a comparative evaluation of K - Nearest Neighbour and Naive - Bayes classifiers to improve prediction of short-term precipitation in Numerical Weather Prediction (NWP) models. Aiming at providing alternatives to imperfections under the current NWP physical process methods namely representation of many points of the earth at high resolution, the dynamic nature of the atmosphere, errors in the initial conditions, inefficiency due to sparsely distributed data points. These short comings require large amounts of computing resources to effectively stimulate and predict weather changes. One of the many unsuccessful interventions is upgrade of hardware components to cope up with the requirements for high computer processing speed, storage capacity and need to continue increasing the resolution at which NWPs run while keeping power consumption with in the reasonable limits. Machine learning KNN and Naive Bayes scalability capabilities enable the NWP model to handle growing amounts of work while keeping computational cost within the reasonable limits. Research studies have demonstrated the capabilities of machine learning models to replace weather and climate models that are based on the physical processes and the basic equations of motion. In this research, input parameters are Maximum and Minimum temperature, Relative humidity, Dewpoint temperature, Dry and Wet bulb temperatures and Rainfall amount six hourly interval observation datasets for 03 (three) years from Uganda National Meteorological Authority. The observation datasets are for Kawanda, Namulonge and Masindi weather stations. These stations are within the same climatic influence of River Mayanja. Measurable metrics being Accuracy, F1 Score and Mathew correlation coefficient (MCC). F1 score and MCC experimental studies offers performance analysis. The data imbalance remedy prone to weather observation datasets was addressed using the Synthetic Minority Over Sampling Techniques (SMOTE) algorithm. Research findings have found the Naive Bayes robust in the prediction of short-term precipitation for both dry and wet season. KNN has high accuracy, low F1 score and MCC. This research asserts the KNN being weak in generation of independent datasets, thus not fit to be used in a weather prediction problem. These research findings can be taken on further into implementation of scalable NWPs for weather and climate products using minimal computation resources, lowering the overall budget constrains for National weather services across the globe, ultimately increasing access to accurate weather forecast, accelerating growth of other sectors of the economy like Agriculture, Transport, Construction, Manufacturing and Tourism.
  • Item
    Designing persuasive technologies for societal benefit: a persuasive technology for fighting electricity theft in Kampala, Uganda
    (Makerere University, 2021-07) Mbabazi-Mutebi, Ruth
    Electricity theft is a major challenge for electricity utilities world over, not sparing Umeme, Uganda. Although, the utility invests heavily into electricity theft reduction measures, some of these measures are frequently breached and progress is slow. These electricity theft reduction efforts can be aided by persuasive technologies. Persuasive technologies are a promising mechanism for attitude and behaviour change. They are being developed to handle a plethora of challenges ranging from health, safety, financial management, to energy conservation. However, while a lot of attention has gone into developing these technologies, little has been given to the theoretical work of developing persuasive technology design frameworks. Existing frameworks have limitations that make it difficult to utilize them in solving societal problems. The aim of this research, therefore, was to propose a persuasive design framework for developing persuasive technologies for societal benefit focusing on electricity theft reduction in Kampala, Uganda, as a case study. Hevner’s design science research methodology, was employed in an action research approach to modify Fogg’s eight step process, one of the existing persuasive technology design frameworks. As input and justification for the research, we studied electricity theft in Kampala and Cape Town using survey questionnaire among electricity consumers and interviews with electricity utility staff. The study found that despite using split prepaid meter, electricity theft still prevails in Cape Town. The study in Kampala revealed that 68% of electricity consumers were not willing to participate in fighting electricity theft. This justified the need for a persuasive technology intervention. A literature review of persuasive technologies for societal benefit revealed that Fogg’s eight steps is the best framework to use, though it lacked guidelines on how to propose requirements. It was modified using design theory resulting into the Design Theory-Fogg’s Eight Step Process (DT-FESP). It was used to propose requirements and persuasive techniques for a mobile Android application to increase willingness to participate in fighting electricity theft called, “Faayo” (available at www.faayo.net). The requirements upon which the application was built where validated through; a) interviews, b) survey questionnaires, c) field studies, and d) laboratory studies. Participants were; a) Umeme staff, b) psychology experts, and c) electricity consumers. Overall, 84.6 % of requirements were found to be potentially persuasive. The overall mean for all requirements was 3.9 (using a Likert scale of 1 for strongly disagree, 3 for neutral and 5 for strongly agree). This validated the DT-FESP framework used to propose them. The research demonstrated the role and feasibility of persuasive technologies in reducing electricity theft and recommended that Umeme includes persuasive technologies in their electricity theft mitigation strategies. It also proposed a design framework for developing persuasive technologies for electricity theft which is a benefit to society.
  • Item
    Detecting money laundering using a pattern matching approach based on Rete Algorithm
    (Makerere University, 2020-09) Kiwanuka, Patrick Ivan
    Money laundering is the criminal practice of filtering funds gotten from illegal activities through a series of transactions so that the funds appear as proceeds from legal activities. Money laundering is a diverse, complex process and basically involves three independent steps namely placement - placing, layering, and integration. It mainly occurs within financial institutions and it is difficult to detect [1]. All financial institutions in Uganda are mandated by the Central Bank to have mechanisms in place for detecting money laundering activities. The Parliament of Uganda passed an Anti-Money Laundering Act Law in 2013 which criminalizes these activities. In addition to the above laws, many institutions have developed policies internally that help them in detecting money laundering activities. Money laundering can have negative effects on the economy of the countries such as loss of government tax revenue, driving up the cost of government due to the need for increased law enforcement to fight the vice, undermining the legitimate private sector and integrity of the financial markets since Money launderers often use front companies which co-mingle the proceeds of illicit with genuine ones. These front companies have access to substantial illicit funds, allowing them to subsidize front company products and services at levels well below market rates. The objective of this project was to develop and validate money laundering detection rules using rete pattern matching algorithm. Through case studies with core banking systems of two local banks and related systems, requirements were established to inform the design of the patterns linked to potential money laundering activities. These patterns informed the design of rules such as the Daily transactions limit and extended the existing rules with a dynamic scoring system where a threshold score combines a net laundering score at run time. A banking system prototype was developed using python and SQLite for the database. The designed rules were implemented using Jess rule engine and integrated into the prototype banking core system. The system integrates dynamic rules making it hard for criminals to deduce the transactional limits. A scoring system has been introduced which evaluates each transaction against all the rules and categorizes it based on the overall score. This reduces on the number of false positives. Jess wrapped in Java was used to develop the improved rule sets and a scoring system that detected suspicious transaction patterns. Several transactions were carried out using the banking applications and suspicious transactions were highlighted by the system.
  • Item
    Detection of WannaCry Ransomware using machine learning techniques
    (Makerere University, 2022-03-16) Opio, Arthur Moses
    As the modern society embraces the digital age, there are powerful threats like malware that are developing daily and they continue to impact a large number of computing devices. Malware is malicious software that are designed to cause harm as intended by the malicious actor. Today’s ransomware families implement very sophisticated encryption, obfuscation and propagation schemes that limit the ability to recover the lost data, even if the ransom is paid, there is no guarantee. Security researchers continue to use the signature-based and behavioral based detection but that is not enough. We collect the data, preprocess, perform feature extraction and build the classifiers that are applied to the various supervised machine learning algorithms with the mode. We built an artificial intelligence model to detect wannacry ransomware using the machine learning classification algorithms. We present our ransomware analysis results on both the static and dynamic analysis and our developed machine learning model. To prove our concept, We used the wannacry dataset together with other two malware datasets to train and test the performance of the various classification algorithms. The datasets were explored, pre-processed, and split into training data and testing data with a ratio of 7:3. During data collection, we ensured to obtain good training data. This resulted into good machine learning classifiers for Random Forest, Gradient Boost and KNN with a performance of 99%. We also used the Deep Neural Multilayer Perceptron algorithm which also had a performance of 98%. With these results, this shows machine learning can be used to detect wannacry on infected machines and prevent it from spreading.
  • Item
    Developing a framework for measuring IT project success in developing countries : a case study of Uganda
    (Makerere University, 2021) Barigye, Innocent
    Over the past years, measuring project success had been concentrated on the performance of project management more especially on cost, time and quality but not considering product operation and involvement of the clients and the employees. In Organizations and Companies, Successes on a project means that certain expectations for a given participant were met, whether owner, planner, engineer, contractor or operator. However, these expectations may be different for each participant and the study of project success and project frameworks is often considered as one of the vital ways to improve the effectiveness of project delivery (Chan et al., 2004). It is reasonable to believe that, if we meet the quality, time and cost targets for a project, it will be considered successful. Apparently, there are other factors that stakeholders deem to be important in determining project success. If Project implementation methodology adopts other relevant criteria for successful projects, organizations/companies would experience improved project execution and delivery. The study used the National Information Technology Authority Uganda (NITA-U) and other IT project management firms like Government (UCC), Airlines, Banking, SMEs in order to widen the scope for our research focusing on IT project management. The main objective of this research is to develop a framework for measuring IT project success, the specific objectives for the study were, To determine the factors for IT project failure and success in the developed framework for developing countries, To identify the appropriate measures that can be used to achieve the IT project success, To assess the relationship between the framework criteria and IT project success in developing countries. The research design used here was a cross-sectional research design and involved quantitative and qualitative approach to data collection but also literature review approach was used to compare several project frameworks. The survey results showed that Time (74.4%), Cost (55.6%) and Quality (51.1%) still remain the important criteria for assessing the performance of IT projects in the minds of professionals. However, the users seemed to consider Time and Cost more important than Quality. Project should also target at satisfying the needs of the key stakeholders. The key stakeholders for an IT project are: the software users’ satisfaction (77.8%), and the customer satisfaction which stood at 55%% from the data got from the respondents for this study. From both the survey and the literature reviewed the researcher has made modifications to the four basic perspectives based on the following views: The I.T. projects are commonly carried out for the benefit of both customers and software users. Project personnel and project team should be as internal customers that should benefit from the project. If we give more attention to human resources, we will then have an excellent basis for improved results in the rest of the perspectives. From the study several measures were suggested in order to achieve the project success like good planning, adequate resources, Effective communication, Cost and time measures of control Deadlines & budgeting, Good risk management. Any project is only good if it is functional. Nothing else matters much if for example a software program is not accepted by its users. Therefore, having a clear definition of project success as the establishment of a set of success criteria is of utmost importance for every project-oriented organization. If an organization does not know early on the project how they are going to measure its business success, they will surely be faced with unpleasant situations in the long run.
  • Item
    Electricity theft in Kampala and potential ICT solutions
    (SPRINGER LNICST 147, 2014) Mbabazi-Mutebi, Ruth ; Sansa-Otim, Julianne ; Sebitosi, Ben ; Okou, Richard
    Electricity theft is the main source of non-technical losses in electricity distribution utilities. This paper presents data from an ongoing research to study the causes of electricity theft in Kampala, Uganda and people’s response to the efforts being made to reduce it. Our study reveals that electricity theft in Kampala is largely due to economic reasons and corruption within the utility company. It confirms that people perceive electricity theft as the utility’s problem and are not willing to report electric theft suspects. We propose ICT technologies to encourage consumer participation in reducing electricity theft.
  • Item
    Enhancing the DCFM node isolation attack mechanism for OLSR protocol in Android MANETs
    (Makerere University, 2017-11-23) Kawulira, Edwin
    Mobile Ad hoc Networks (MANETs) have recently gained wide adoption by their ability for communication amongst users without infrastructure for instance Wireless Mesh Community Networks that are setup not to rely on telecommunication infrastructure because they may be too expensive, damaged from natural disasters or simply nonexistent. However, major investigations have mainly focused on routing protocol problems with little progress in solving secure routing in MANETs. This in turn has led to the proliferation of threats and vulnerabilities like the Node Isolation attack against Optimized Link State Routing - OLSR one of the most widely used MANET protocols where a malicious node attacks by exploiting topological knowledge of the network to isolate the victim from the rest of the network and subsequently deny communication services to the victim. This project adopts the Denial Contradictions with Fictitious Node Mechanism (DCFM) which we enhance with Group Testing techniques that yield better and more efficient detection rates against node isolation attacks by employing the same tactics used by the attacker itself. This DCFM enhancement is achieved through modelling and construction of a Colored Petri Net (CPN) model of the mandatory parts of the OLSR protocol for formal verification of its behavioral correctness. The applications of Colored Petri Nets and state space analysis tool have been successful in modelling and performing analyses of the OLSR protocol with DCFM demonstrated success metrics of increase in detection rates of over 95 percent of attacks and a very high reduction in delay latency attributed to Group Testing’s disjunct matrices techniques and finally after demonstrating how the construction of executable formal models such as a CPN model can be a very effective way of systematically reviewing an industrial-size protocol specification for security verification and formal behavioral analysis which can be employed to other security attacks.
  • Item
    Enhancing the interoperability of public health systems in Uganda using a data exchange module
    (Makerere University, 2017-10) Sekiwere, Samuel
    With an increasing adoption of digital health systems by the Ministry of Health in Uganda, adequate digital health systems interoperability has become a major concern. The need for these systems to coexist and complement each other has also increased the need for 1) secure and reliable health information exchange, 2) scalable message transfer and 3) simpli ed management and control for information being exchanged. In our project, we have developed a secure data exchange module (Dispatcher2) that can be used to integrate at least two disparate digital health systems. Dispatcher2 implements a simple commu- nication protocol that governs information exchange between any pair of applications it integrates. It relies on a persistent queue to o er message persistence and improved traceability for the information being exchanged. It also ships with management web interfaces that manage the message queue and con gurations for the systems it integrates. Our project has endeavored to follow existing interoperability standards and recommendations whenever possible. It has been simulated with 3 publically used health systems in Uganda, that is; mTrac, mTracPro and DHIS 2. The performance of the solution has also been analyzed and it was able to achieve a throughput of over 300 transactions per second and an average response time of about 9.78ms on laptop hardware. More results from our project are discussed in detail in this report.
  • Item
    Extraction content recommender model for a personalised e-learning environment based on learner’s course assessment feeds
    (Makerere University, 2022-11-24) Twijukye, Brian
    E-learning, also known as online learning or electronic learning, is the acquisition of knowledge through digital technologies and media. In most cases, it refers to a course, program or degree delivered completely online. There are many terms used to describe learning that is delivered online, via the internet, ranging from Distance Education, to computerized electronic learning, online learning, internet learning and many others. Most authors point that consideration of the learner profile (personality, preferences, knowledge, etc.), is an essential and an important element in achieving an efficient and successful teaching. Therefore, it is extremely delicate and difficult to achieve a personalized learning scenario for each learner in the traditional closed classroom. Searching and retrieving information on E-learning environment is inconvenient, inefficient and sometimes time consuming as it relays irrelevant information that requires much time for student to scrutinize it and make meaning out of it, and in this case there is no previous work that has covered how learner’s assessment feeds like uploaded file’s content while answering E-learning environment set course works and quizzes can be incorporated in the extraction content recommender model to create a personalised E-learning environment. The main purpose of this work was to develop a an extraction content recommender model for e-learning personalisation based on learner’s course assessment feeds, which will allow students to obtain results according to their profiles and interests without taking too much time on the E-learning environment while searching relevant information. And the simulation was conducted on the extraction content recommender model with three datasets user profile ,learner's course assessment feeds and E-learning domain resources, we found out that using assessments feeds with different topic discussions in it recommends the value with highest rates and more recommendations are observed on it and when it comes to performance the learner's course assessment feeds with more words and one topic discussion takes too much time to recommend compared to learner's course assessment feeds with few words and more topics discussion. In the future work will be able to enable of our model to extract more assessment feeds from more than one section of the learner uploaded pdf assessment feeds and filter extracted information to get more key words of different topic’s discussions.
  • Item
    Fracture detection in children using convolution neural network
    (Makerere University, 2019-07) Nanziri, Eunice
    Fractures in children are among the medical cases that take a lot of doctors’ attention and time while analyzing the images to identify the presence or absence of fracture. More so a tired doctor may fail to identify the presence of fracture after looking at many images with health bones which may result in poor treatment. Our method of fracture can help doctors in faster detection, meet their deadline but also can help in obtaining accurate results hence administer proper treatment. In this research, we used X-ray images which we obtained from Mengo hospital. We obtained 60,140 Images and with the help of a radiologist we were able to find 211 images of children from the entire data set and also, we were able to separate images with fracture from those with fracture. we split our data into 3 different sets (train set, validation set, and test set). We used a convolution neural network (CNN) a method of automatic detection that is mostly preferred for cases that require deep analysis. We trained our model on 211 x-ray images and used a kernel to generate different features of interest. we obtained 4 convolution layers while developing our model and a max-pooling layer was placed in between each 2d convolutions layers to record the weights of the parameters but also to reduce over-fitting. We used 3 different matrices to evaluate our model which include accuracy, Precision, and recall. On training the model, we obtained Test accuracy levels of 47%, 59% for test precision, and Test recall of 76%. while using our data-set and we noticed an over-fit because of the small data set used, we then applied data augmentation and obtained an accuracy level of 56%, test precision 49%, and test recall of 76%. We also trained our model on VGG Net a model and we obtained an accuracy level of 63%, test precision of 66%, and test recall of 79%.
  • Item
    A framework for GIS-enabled public e-participation in municipal solid waste management
    (Makerere University, 2020-07) Arinaitwe, Irene
    Municipal solid waste is a serious environmental challenge that affects most urban authorities globally. Annually, more than two billion metric tons of wastes are generated globally. Municipal solid waste is especially a serious challenge to many developing countries because of a high waste generation stemming from high population growth and rapid urbanisation. In Sub-Saharan Africa alone, at least 62 million tons of waste are generated annually. The high rate of waste generation has made municipal solid waste management (MSWM) to be the single most crucial function of urban authorities in most developing counties, including Uganda. However, most urban authorities cannot cope with the high demand for MSWM because of weak infrastructure, limited funding, and weak legal and regulatory framework. Additionally, there is limited public engagement in MSWM amidst prevalent poor attitude towards waste management initiatives, by the public. Public participation is central to the achievement of sustainable waste management systems. However, there is minimal public participation in government administrative processes, including municipal solid waste management, in many sub-Saharan countries. Limited public participation in MSWM is attributed to the lack of effective public participatory platforms that ensure inclusive and broader stakeholder engagement. Although public participatory geographic information systems (PPGIS) have the potential to enhance public participation, there are inadequacies in theoretical foundations to guide their implementation. Therefore, geographical information systems (GIS) tools that enhance broader and effective stakeholder engagement have not been leveraged to address the challenges of MSWM in most developing countries. The implementation of PPGIS requires theoretical frameworks that support systematic analysis of not only the technical aspects but also the social-behavioural principles. Although several frameworks for public participation exist, they are also not tailored to address the contextual barriers to public participation, especially in the developing-country context. Therefore, this research aimed at developing a framework for GIS-enabled public e-participation in MSWM (coined as GPEP) in urban authorities, especially in developing countries. Pragmatism was selected as the philosophical underpinning for this research because this research aimed at finding a practical solution that is contextually suited to address MSWM challenges. The abductive approach research approach was used to maximize the strengths of both deductive and inductive research approaches. Pragmatism and the abductive approach would ensure a comprehensive yet contextually suited framework. In the development of GPEP, the design science research method was used. Design science research method supports the development of tangible and practical solutions to societal problems, including public eparticipation in MSWM. GPEP extended the Enhanced Adaptive Structuration Theory (EAST-2) that was developed by Jankowski and Nyerges to include constructs and aspects from Adaptive Structuration Theory (AST), Enhanced Adaptive Structuration Theory (EAST), Dynamic capabilities theory, and the institutional analysis and development framework. Also, GPEP includes factors that influence the uptake and use of PPGIS applications identified from the literature and contextual requirements derived from an exploratory study that was conducted in the greater Kampala area. GPEP was tested with a descriptive field study. Only the constructs that met the validity and reliability thresholds were included in GPEP. Data were analysed using the partial least squares (PLS) technique of structured equation modelling (SEM). The analysis involved an iterative stepwise forward factor selection process whilst determine the relationships between aspects and constructs. A two-tailed statistical significance of 0.05 was used and power set at 80%. The analysis found that technology influences and task influences have a direct bearing on conducting GIS-enabled public e-participatory processes (p<0.001 for all). Similarly, Social Institutional influence and participant influence had insignificant influence on GIS-enabled public e-participatory processes, contrary to the findings from prior research. GPEP was evaluated using structured walkthroughs and experimentation methods. GPEP was evaluated on a five-point Likert based on its Usability, Feasibility, Completeness, and Consistency attributes. The evaluation reported high usability scores (mean of the mean scores=3.7), feasibility (mean of the mean scores=3.9), and completeness (mean of the mean scores=3.70), and consistency (mean of the mean scores =4.4). This research achieved three objectives. (1) The requirements for GPEP were determined, (2) GPEP was designed, and (3) GPEP was evaluated to ascertain its practical utility. GPEP contributes to theory in three ways, (1) the factors that influence the adoption and implementation of PPGIS were identified, (2) a combination of methods-structured walkthroughs, experimentation and structured equation modelling were used evaluate the GPEP, and (3) the GPEP was derived through the integration of aspects and constructs from different theoretical frameworks. The study findings point to the need for a proper understanding of participant characteristics, consideration of organisational workflows and business processes, and task technology fit assessment before applying PPGIS in environmental public participation. Additionally, the implementation must be preceded by change management, feasibility studies, resource mobilisation, public awareness, and GIS infrastructure investments. Overall, GPEP should be used to inform the implementation of a PPGIS in Uganda. This study was cross-sectional, and the long-term benefits of this research cannot be determined at this stage, and this should be determined using a longitudinal study.
  • Item
    A framework for information management in E-agriculture
    (Makerere University, 2022-12-20) Mugejjera, Emmanuel
    Agriculture is a vital sector in a developing economy like Uganda’s. ICTs have been used in this sector to avail information and to support different information based agricultural processes in what is called electronic agriculture. Despite the use of ICTs, access to agricultural advisory information in a developing economy like Uganda’s remains problematic. This state of affairs is attributed to inadequate management of agricultural advisory information in e-agriculture. Therefore, this study aimed to develop a framework for supporting management of agricultural advisory information for small scale farmers engaged in growing of crops aided by ICTs in Uganda’s developing economy. The Design Science research method was used to guide the development of this framework. The framework presented in this work was based on a field study using 386 respondents from Uganda’s districts of Gulu, Lira, Mbale, Namayingo, Masaka, Wakiso, Mbarara and Ntungamo. Structural equation modeling was used in the design of the framework. The results show that the critical success factors for management of agricultural advisory information are: People and Technology; Funding, Processes, and Regulations; and Information use outcomes and continuity. The framework is composed of the above factors with People and Technology; Funding, Processes, and Regulations; influencing Information use outcomes and continuity. The framework was evaluated by seeking expert opinion and using a prototype in form of a web-based platform. In conclusion, the findings indicate that the framework is suitable for supporting the management of agricultural advisory information based on the parameters of goal, environment, structure, activity and evolution. It is suggested that the framework be used based on practical suggestions provided on each sub factor of the framework to aid policy makers in information management in e-agriculture support the agricultural advisory information management practices. Overall, the framework can be used to inform the management of agricultural advisory information. The prototype developed is a foundation for automation of selected tasks in the management of information in e-agriculture in Uganda’s context.
  • Item
    Hybridizing machine learning and static malware detection using the PE header
    (Makerere University, 2021-11-23) Kipsang, Jacob
    Cyber crime cases currently involve demanding payment after infecting a victimized organization’s computers with ransomware or impairing operations through a distributed denial-of-service attack which significantly impacts the confidentiality, integrity and availability of data. Recent researchers show that hybridizing techniques can detect malware or benign effectively. Our research provides an experimental study on hybridizing machine learning and signature-based techniques to detect malware based on the PE header information. The dataset was sliced randomly into training 80% and testing 20% sets. The classifiers we used were Random Forest, Gradient Boosting and Ada boost to train and test the dataset. We evaluated our models using the evaluation metrics. Results showed overall achieved accuracy is high for the cleaned dataset ranging from 99.70% to 99.77%, for the uncleaned dataset range from 93.83% to 96.83%. The VirusTotal file report API had a high Average detection rate for unclean datasets ranging from 0.00% to 12.57% and a low average detection rate of 0.00% on a cleaned dataset. Random Forest emerged as the best classifier for both cleaned and uncleaned datasets with an average detection rate for static analysis of 0.00%.
  • Item
    Information seeking behaviour of rural women in Bungokho County, Mbale District of Uganda
    (Makerere University, 2019-12) Khanyalano, Justine
    The aim of the study was to investigate the information seeking behaviour of rural women involved in economic activities in Bungokho County, Mbale District in order to suggest strategies to enhance the information seeking. The following objectives guided the study; to examine the information needs of rural women of Bungokho County, to identify the information sources/resources used by rural women of Bungokho County to seek and satisfy their information needs, to establish the challenges faced by rural women of Bungokho County in seeking information and to suggest strategies to enhance the information seeking of rural women of Bungokho County. Mixed-Methods approach was used with survey design for quantitative and case study for qualitative. Questionnaires and focus group discussion were used to collect data. Data was analysed using thematic analysis and Statistical Package for Social Scientists (SPSS). The findings revealed that respondents’ commonly needed information regarding pests and disease management followed by income generating activities. The rural women consulted informal than formal sources of information and social-political meetings like burial ceremonies, weddings and harvest festivals were the most used channel of communication. The major challenges that respondents faced was lack of awareness and knowledge on where to get useful information and financial obstacles. The strategy seconded by the respondents was exposing and training the rural women in different income generating activities and women empowerment through education and enlightenment programs. The study recommended regular training of women in income generating activities, financial literacy programmes, making loan facilities available to the rural women for business investments, repackaging of information, introduction of adult education in different villages, deployment of extension workers and gender sensitization.
  • Item
    An information technology based framework for integrating referral levels among public healthcare facilities in Uganda
    (Makerere University, 2021-12) Nakayuki, Mildred
    An effective referral system ensures a close relationship across all levels of the healthcare system to ensure that people receive the best possible healthcare. Coordination among referral levels can be achieved by strengthening communication and feedback between the different levels of healthcare facilities, supervisors, and healthcare professionals based in their respective healthcare facilities. This can improve the access of patient management details and better services provided to patients at all times hence promoting efficiency and effectiveness in the referral system (Heeringa, Mutti, Furukawa, Lechner, Maurer & Rich, 2020). This study investigated the challenges faced by referral levels such as lack of coordination, limited patient management details transferred with the patients across various levels, among others to design an Information Technology (IT) based framework for integrating referral levels among public healthcare facilities in Uganda. This study used the design science research methodology with the output of an artefact which is the designed framework of integration. An exploratory study was conducted to collect primary data among 84 respondents (76 questionnaires and 8 interviews) from four selected public healthcare facilities at various levels of healthcare which included Health Centre, General Hospital, and National Referral Hospital. The data collected was analysed and challenges faced by referral levels were revealed where requirements for designing the framework were derived from and the existing frameworks that were reviewed. The framework was tested and validated following design science parameters (efficiency and usability). Results from testing and validation indicated that the framework addressed the challenges facing referral levels and achieved the objective of the study. Results from this study indicated that 100% of referral levels in the healthcare facilities either make or receive referrals and the majority of these respondents agreed that it is important for referral levels to be integrated and these respondents constituted 92.4% while 7.6% disagreed that it is not important to integrate referral levels. The referral system in Uganda faces challenges with a 97.3% level of agreement and the greatest challenge was the failure to capture patient referral data and keep track of these patient records at some levels of healthcare facilities with an average mean of 4.857. Findings revealed that the majority of the respondents agreed that it is important for referral levels to be integrated and these respondents constituted 92.4% and 7.6% disagreed that it is not important to integrate referral levels. And the greatest importance of integration towards solving the challenges affecting the referral system was integration improves and streamlines communication among healthcare providers involved in a patient’s care across different levels of healthcare facilities with a mean of 4.799.