PhD Graduands

Abstract: Presidential campaign periods are a major trigger event for hate speech on social media in almost every country. A systematic review of previous studies indicates inadequate publicly available annotated datasets and hardly any evidence of theoretical underpinning for the annotation schemes used for hate speech identification. This situation stifles the development of empirically useful data for research, especially in supervised machine learning. This paper describes the methodology that was used to develop a multidimensional hate speech framework based on the duplex theory of hate [1] components that include distance, passion, commitment to hate, and hate as a story. Subsequently, an annotation scheme based on the framework was used to annotate a random sample of ~51k tweets from ~400k tweets that were collected during the August and October 2017 presidential campaign period in Kenya. This resulted in a goldstandard codeswitched dataset that could be used for comparative and empirical studies in supervised machine learning. The resulting classifiers trained on this dataset could be used to provide real-time monitoring of hate speech spikes on social media and inform data-driven decision-making by relevant security agencies in government.

Abstract: 

An unbiased ratio-type estimator has been considered by Oyoo et al. [8] in estimating finite population total under non-response in stratified random sampling. In their paper, Oyoo et al. [8] assumes that variation in the ratio   is much higher than variation in the response variable, Y and in the auxiliary variable, X and therefore, while obtaining the expression for the variance, variations in Y and X are considered negligible. In this paper, however, we consider variations in R, Y and X and use the result by Hartley and Ross [3] and Goodman and Hartley [9] to obtain the expression for variance the estimator by Oyoo et al. [8]. We examine performance of this estimator using simulated data and the results show that this estimator performs better than the usual (biased) ratio estimator constructed in simple random sampling without replacement (SRSWOR) and under stratified random sampling.

Keywords and phrases:

unbiased ratio estimator, mean square error (MSE), non-response.

Abstract:

Air samples were collected from three urban and one rural sites in Kenya with the aim of establishing pollution levels of Polybrominated Diphenyl Ethers. Forty-eight air Samples were collected by passive air sampling, Soxhlet extracted and analysed for brominated diphenyl ethers using gas chromatography coupled with mass spectrometer. The mean concentration of polybrominated diphenyl ethers residue in air ranged from ≤0.9 to 152.72±3.19 pgm−3. The predominant congener was 2,2′,4,4′-tetra-bromodiphenyl ether with mean concentration range of 1.94±0.03 to 152.72±3.19 pgm−3 followed by 2,2′,4,4′,5-penta- bromodiphenyl ether with mean concentration range of 1.32±0.06 to 66.83±1.19 pgm−3. Seasonal variations of the pollutants showed a high level of Polybrominated Diphenyl Ethers in hot dry season in range of 1.94±0.03 to 152.72±3.19 pgm−3. Air samples from Dandora and Industrial area both from urban location recorded high concentrations of the analysed polybrominated diphenyl ethers compared with the air samples from the rural location.

Abstract: 

This study investigated spatial and temporal trends of rainfall and temperature in the Amboseli ecosystem of

Kenya. The analysis were based on historical Climate Hazards group InfraRed Precipitation with Station (CHIRPs) and Climate Hazards group InfraRed Temperature with Station (CHIRTs) data for the period 1960-2014 and the period 2006-2100 for the projections. This data was used due to limitations in the observed station data. Projections of rainfall and temperature were based on Regional Climate Models (RCM) from Coordinated Regional Downscaling Experiment (CORDEX) over the Amboseli ecosystem. The long-term annual and seasonal trends of rainfall and temperature were analyzed via Mann– Kendall’s statistical test and linear trend analysis. The annual and seasonal rainfall declined slightly between 1960 and 2014 though not significant. However the temperatures increased more in the annual minimum (1.23 °C) compared to the annual maximum (0.79 °C). The maximum temperatures for the October-November-December (OND) season had highest increases of 0.88 °C while the March-April-May (MAM) season showed an increase of 0.69 °C. The highest increase in minimum temperatures of 1.35 °C was recorded for the June-July-August-September season (JJAS), while the least increase was in MAM (1.04°C). Projected rainfall based on Representative Concentration Pathways (RCPs) for the periods 2006-2100 varied with RCP 2.6 showing a decline for the four seasons. RCP 4.5 and 8.5 project marginal increase in annual and OND with declines in the MAM and JJAS. Projected maximum and minimum temperature for RCP 2.6 indicate increments of less than 1 °C while for RCP 4.5 the maximum range is between 0.57 °C and 1.85 °Cand minimum is between 0.51 °C to 1.98 °C. RCP 8.5 projected maximum increase are the highest between 1.11°C and 4.34 °C and minimum is between 1.34 °C and 5.26 °C based on period – 2030, 2050 and 2070. The increase of temperatures and changes in rainfall can have large impacts on the resources in the savanna dry lands of East Africa especially on its livestock, agriculture, wildlife and pastoral and agro-pastoral communities.

Index Terms— Climate change, dry lands, mitigation,Representative Concentration Pathways, temperature and

rainfall

Abstract

Entomopathogenic nematodes (EPNs) are worldwide soil-dwelling insect parasitic nematodes. They are potential pest bio-control agents a key component of Integrated Pest Management (IPM) programs. This study aimed to characterize and evaluate the pathogenicity of an EPN isolate from Kenya. The nematode was isolated from soils using insect bait technique and both morphological and molecular identification was performed. Efficacy of the isolate was evaluated against Tomato leafminer larvae (Tuta absoluta Meyrick.) using dose-based treatments of 0-control, 100, 150, 200, and 250 infective juveniles (IJs/ml). Morphological analysis revealed body length (L) of 835(659-987) µm and 1781 (1297-2097) µm from fresh IJs and males respectively. Males lacked a mucron. The isolate was characterized by the partial sequence length of 877 bp of the ITS region. Blastn results indicated the EPN isolate had a similarity match of 81-92% with Afro-tropical Steinernema species. It matched with Steinernema sp. (AY230186.1) from Kenya at 92% and Sri Lanka (AY230184.1). Phylogenetic analysis placed the isolate together with Steinernema sp. (AY230186.1) and (AY230184.1) with a bootstrap value of 100%. Maximum mean larval mortality (80%; 96%) was achieved 24 and 48 h post-treatment at concentration 150 IJs/ml. All nematode concentrations achieved over 50% mean mortality after 24 h period. There was a significant difference (P = 0.001) between doses 150 and 200 IJs/ml. From the study, it was concluded that the nematode isolate was Steinernema sp now referred to as Steinernema sp. Kalro (Genebank Accession MW151701). The EPN has the potential for development as a biological control agent against T. absoluta.