Browsing by Author "Kirumbu, Michael Kiura"

Now showing 1 - 4 of 4
  • Thumbnail Image
    Item
    A Hidden Markov Model of Risk Classification among the Low Income Earners
    (Journal of Finance and Economics, 2018-12-18) Ntwiga, Davis Bundi; Ogutu, Carolyne; Kirumbu, Michael Kiura; Weke, Patrick
    Low income earners have volatile incomes and most financial providers shun this group of borrowers even though they are motivated in managing the limited resources they have through savings and investments as a means to lower the fluctuations of their income. Peer groupings of the low income earners can assist in pooling the resources they have and improve the group risk mitigation process as group members act like social collateral in credit lending. The study used Kenya Kenya Financial Diaries data of 2013 from 280 households to analyze and understand the credit quality levels and credit scores of peer groups versus individuals among men and women. Hidden Markov model classified the low income earners into credit risk profiles wih a view of understanding the role of groups in low income group lending. Peer groups diversify risk inherent in individual borrowers with women only groups having higher credit quality levels as compared to men only groups. Women and their respective peer groups are more stable with less variability as compared to men. Financial technology providers can incorporate the wide array of soft information to lend to low income earners through mobile based peer groups.
  • Thumbnail Image
    Item
    Inclusion of peer group and individual low-income earners in M-Shwari micro-credit lending: a hidden Markov model approach
    (International Journal of Electronic Finance, 2018-04-10) Ntwiga, Davis Bundi; Ogutu, Carolyne; Kirumbu, Michael Kiura
    The M-Shwari micro-credit lending system has excluded the low income earners as they lack good financial options due to volatile and fluctuating income. This paper proposes a decision support system for credit scoring and lending of the low income earners who are customers of M-Shwari using the hidden Markov model. The model emits the credit scores of the customers, both for the peer groups and the individual customers. The learning and training of the model utilises the customers' socio-demographics, telecommunication characteristics and account activities. The peer groups have higher credit scores and are more attractive to offer credit facilities using M-Shwari when compared to the individual borrowers.
  • Thumbnail Image
    Item
    Trust Model for Social Network Using Singular Value Decomposition
    (Interdisciplinary Description of Complex Systems, 2016-03-18) Ntwiga, Davis Bundi; Weke, Patrick; Kirumbu, Michael Kiura
    For effective interactions to take place in a social network, trust is important. We model trust of agents using the peer to peer reputation ratings in the network that forms a real valued matrix. Singular value decomposition discounts the reputation ratings to estimate the trust levels as trust is the subjective probability of future expectations based on current reputation ratings. Reputation and trust are closely related and singular value decomposition can estimate trust using the real valued matrix of the reputation ratings of the agents in the network. Singular value decomposition is an ideal technique in error elimination when estimating trust from reputation ratings. Reputation estimation of trust is optimal at the discounting of 20 %.
  • Thumbnail Image
    Item
    Use of Objective Tests in Examining Law Courses at Daystar University
    (Paradigm Academic Press, 2024-09) Wekesa, Moni; Mikinyango, Asha; Kirumbu, Michael Kiura; Wandera, Susan N.; Wekesa, K. T
    The use of multiple-choice questions (MCQs) in law schools has not gained widespread acceptance, unlike in medical schools where they enjoy global usage. Law Schools traditionally use essay-type/problem-solving questions to assess students. The efficacy of this form of assessment is increasingly being attacked due to increasing numbers of students and advancements in technology that enable students to generate answers. Bloom’s Taxonomy provides for a hierarchy of learning processes, which include ‘remember’, ‘understand’, ‘apply’, ‘analysis’, ‘evaluate’, and ‘create’. Studies on whether MCQs can test higher-order learning processes required in law courses have been inconclusive. We did a retrospective study to investigate whether MCQs are an efficient and effective way of assessing law courses. Underlying this study was the desire to find an alternative mode of assessment to overcome the threats facing the essay type. Results from selected law courses were analyzed in which students’ performance on MCQ tests, oral tests, and final examinations were compared. MCQs were analyzed on a two-dimensional Bloom’s table to establish the extent to which they tested higher-order learning processes. We compared the results of scores on the MCQs with those of oral tests and final essay-type examinations using correlational analysis and one-way analysis of variance (ANOVA). This was for the courses Constitutional Law (n = 22, MCQs = 75), Intellectual Property Rights (n = 113, MCQs = 100), Broadcast & TV Law (n = 11, MCQs = 76), Administrative Law (n = 65, MCQs = 91), and Cyberspace Law (n = 28, MCQs = 101). In general, students performed best on the MCQ test compared to orals and final exams. A two-tailed correlation analysis comparing all five courses showed a strong correlation between MCQs and Orals (r = 0.699, p = 0.189) and a weak correlation between MCQ test and final exam (r = 0.196, p = 0.752). Sixty-eight percent of 75 MCQs in Constitutional Law tested higher-order processes. The mean scores for MCQ (𝑥̅+𝜎 = 39.45 + 4.83), orals (𝑥̅+𝜎 = 7.64+18.72), and final exam (𝑥̅+𝜎 = 28.5 + 8.88) showed a best performance for MCGs. ANOVA comparing test scores for MCQs, Orals and Final examination showed a very significant difference (F (2, 63, 0.05) = 38.11, p < 0.0001). Fifty-four out of 100 MCQs for intellectual property Law tested higher-order learning processes. The mean scores for MCQ (𝑥̅+𝜎 = 39.78 + 5.22), orals (𝑥̅+𝜎 = 18.88+14.66), and final exam (𝑥̅+𝜎 = 29.0442 + 7.71) showed a best performance for MCQs. ANOVA results were highly significant (F (2,342,0.05) = 125.565, p < 0.0001). In Broadcast & TV Law, 44.7% of 76 MCQs tested higher-order learning processes. The mean scores for MCQ (𝑥̅+𝜎 = 43.27+6.89), orals (𝑥̅+𝜎 = 31.09+12.53), and final exam (𝑥̅+𝜎 = 32.09 + 6.69) were different. ANOVA results were very significant (F (2,30,0.05) = 6.057, p < 0.01). There were 91 MCQs in Administrative Law of which 74% tested higher-order processes. The mean scores for the three tests were MCQ (𝑥̅+𝜎 = 33.05 + 4.76), orals (𝑥̅+𝜎 = 22.89 + 16.08), and final exam (𝑥̅+𝜎 = 32.58 + 5.58). ANOVA results were very significant with F (2,191,0.05)= 20.388, p < 0.0001. Cyberspace Law had 101 MCQs of which 38.6% tested higher-order learning processes.The mean scores for MCQ (𝑥̅+𝜎 = 43.50 + 4.51), orals (𝑥̅+𝜎 = 36.43 + 2.27), and final exam (𝑥̅+𝜎 = 35.04 + 7.71) were different. ANOVA results were very significant (F (2,81,0.05) = 20.375, p < 0.0001). We concluded that MCQs are efficacious and efficient in testing higher-order learning processes. MCQs can be used to assess learning of law courses. We recommend that law schools should embrace MCQs for assessing law courses.