Statistics

Permanent URI for this collection

http://etd.aau.edu.et/handle/123456789/32

Browse

Now showing 1 - 20 of 238

Agricultural Multivariate Stratified Sample Survey with Applications to a Food Supply System
(Addis Abeba university, 1990-06) Zewge, Genene; Melaku, Alemayehu(PhD)
The population studied comprises the. ~gricultur~lhOl~seno.'ds ofa co~uruty in Damota district which is found in the North Omo administrative region. It was stratified by using the agro-ecological factor as the stratifying variable into three strata. Thus, the three strata arrived at in the process were the "Dega", "Kolla", and "Woyna Dega" agroecological zones in the research area. In the case where the study variables of interest are greater than one in a stratified sample survey, the usual optimum (Neyman) allocation method cannot be applied directly since what is best for one study variable may not necessarily be best for the other. Thus, other alternative techruques developed to go about this problem have been compared. These include the compromise, chatterjee as well as proportional allocation. Other methods which involve non-linear programmingtechruques were also reviewed. The data on the study variables which were believed to be indicative of the food supply system of the area were collected These included the family size, acreage, production, expenditure, reserves and purchases of the households. This was done through interviews carried out by going to the sampled households. Prior to the conduct of the final survey, however, a pilot survey was undertaken which covered fifty households from each stratum. This was done in order to obtain the necessary information to estimate the stratum variances which were required for sample allocation purposes. The sample size 2 determination was also done scientifically by similarly using the data from the pilot survey as well as the approved budget of the survey and additional cost estimates. In addition to the above, anthropometric measures of children under five years of age were collected for the purpose of studying the nutritional status of the population \ considered
Analysis of Causes of Cattle Death Using Discrete Regression Models in Ethiopia
(Addis Abeba university, 2015-06) Shiffa, Abdulaziz; Tesfaw, Dejen(PhD)
Analysis of Causes of Cattle Death Using Discrete Regression Models in Ethiopia Abdulaziz Shiffa Addis Ababa University, 2015 Ethiopia is believed to have the largest cattle population in Africa. The cattle sub-sector has been contributing a considerable portion to the economy of the country, food supply and foreign currency income. For the majority of smallholder farmer’s cattle are livelihood and for some farmers the sub-sector is the only livelihood. The objective of this study is to identify the most important factors affecting cattle death in Ethiopia. The dataset in this study is based on the 2014 agricultural survey obtained from Central Statistical Agency (CSA), Ethiopia. Four count models: Poisson, NB, ZIP and ZINB model were applied in order to identify the best fit model for cattle death in Ethiopia. The preliminary predicted values and model selection criteria such as AIC, LRT, deviance and Vuong statistic showed that ZINB regression model is the most appropriate model to fit cattle death. The results of ZINB analysis revealed that age, household size, educational background, farming type, agro-ecology zone and land holding size were found to have statistically significant effect on cattle death. Based on the findings, both governments and stakeholders should work hard to reduce cattle death by promoting mixed farming system and exercising continuous trainings on handling cattle for small holding farmers
Analysis of Climatic Variability and Its Effects on Production of Selected Crop in Ada'a District: Multivariate Time Series Approach
(Addis Abeba university, 2016-06) Tsegaye, Diribsa; Gotu, Butte(PhD)
Climate change affects all economic sectors to some degree, but the agricultural sector is perhaps the most sensitive and vulnerable. In the last three decades, Ethiopia has been affected by climate-related hazards. Agriculture, the most dominant sector in the national economy, has been most at risk because of its dependence on seasonal rainfall. Anticipated climate change has negatively impacted the agricultural sector due to increased temperatures and decreased or greater variability in precipitation, leading to increased food insecurity. This study was carried out with the general objective of examining climate variability and its effects on selected crop production for the last three decades in Ada'a district of East Showa Zone of Oromia Regional State. The relevant data were obtained from the National Meteorological Agency (NMA) and district’s agricultural office. Data on selected crop production, total rainfall and average temperature for the period of 1985 to 2015 were used. The vector autoregressive (VAR) model is employed for modeling. The cointegration relations among the variables were identified by applying Johansen’s cointegration tests, while potential causal relations were examined by employing Granger’s causality tests. Moreover, the short run interactions among the variables were determined through the application of impulse response analysis. The results of the research imply the existence of short term adjustments and long-term dynamics in crop productivity, total rainfall, minimum and maximum temperature. The result of Johansen test indicates the existence of one cointegration relation between the variables. The final result shows that a Vector Error Correction (VEC) model of lag two with one cointegration equations best fits the data
Analysis of Data from Crop Protection Experiments Using Generalized Linear Model: the Case of Parthenium
(Addis Abeba university, 2008-07) Yaregal, Begizew; Taye, Girma (PhD)
Among many weeds that cause crop loss, parthenium was found to be the most terrible one according to some exploratory studies. The problem of parthenium is not only that it cause very sever crop loss, but also it cause health problems to human and animal beings. Control of Parthenium by Farmers ,cultural and labour intensive, caused farmers to suffer from skin allergy, itching, fever, and asthma. This study tried to popularize different Generalized linear models for modeling agricultural data which is used for describing the data sufficiently well and then identify the natural relationship between different variables for further analayis as well as applications. Generalized linear models (GLMs) are used to do regression modeling for non-normal data with a minimum of extra complication compared with normal linear regression. One of the available programs that is important in current statistical practice is the GLM procedure in the SAS software package. The study is based on the result of a parthenium and other species count data, secondary data,obtained from Ethiopian instisute of Agricultre research. Descriptive statistics supported by graphical presentations have been discussed to show the dominance of parthenium on other species per plot area. Furthermore, to evaluate the probability of a plot or a quadrant to be free of parthenium, models form GLM family are applied to the data using SAS software. Based on the parameter estimates, fitted models were formulated, parameters are interpreted and comparison of fitted models conducted. In this model fitting process, an attempt was made to alleviate a confusion of which model to which data. The logit and probit model fitting gives similar results for the same data as expected and the choice of one model cannot be made based on AIC, because the AIC for both models is the same. The poisson regression model fit is found to be inadequate for two different variables, as its Deviance value is far from one. The Negative Binomial Model gives a better fit and its Deviance shows the model is adequate for the same data used for poisson regression. The multinomial logit model for parthenium infestation in five categories as dependent variable and the sum of all other species gives a better result, as infestation level increase i.e as the severity of parthenium infestation increase, the number of the sum of other species gets low which in turn means that the probability of getting other species gets very low
Analysis of Effect of Fine Particulate Matter and Meteorological Factors on Acute Upper Respiratory Infection Hospital Admission in Addis Ababa Ethiopia
(Addis Ababa University, 2022-06-12) Eyasu, Yosef; Temesgen, Shibru (PhD)
Air pollution has become the greatest health concern in our world, especially in respiratory diseases. The concentration of particulate matter of size less than or equal to 2.5 in micrometer (PM2.5) in Addis Ababa is higher than that of the 2021 WHO guidelines limit of the annual average of 5μg/m3 and a daily average of 15 μg/m3 for the last four years of the study period. US EPA controlled single air quality monitor data of PM2.5 concentration for location Addis Ababa-Central is obtained from AirNow.gov and it is used with meteorological data for analysis of health effect. Poisson Generalized Additive Model (GAM) is utilized for the analysis of variables under study from January 1, 2018, to December 31, 2021, in Addis Ababa, Ethiopia. In this study, statistically significant association between fine particulate matter (PM2.5) and acute upper respiratory infection hospital admission in Addis Ababa city and its health effect is observed. Relative risk of acute upper respiratory infection hospital admission associated with 10 μg/m3 increase in PM2.5 concentration was 1.08 (95% CI: 1.06-1.11). We have also observed a positive effect of relative humidity and precipitation on respiratory infection. Therefore reducing the pollutant concentration in the city needs due attention to help people from difficulty in breathing.
Analysis of Electricity Demand in Ethiopia Using Partial Adjustment and ARIMA Modeling
(Addis Ababa University, 2011-05) Abebaw, Misganew; Gotu, Butte (PhD)
This paper examined demand for electricity in Ethiopia as a function of real gross domestic product per capita, real price of electricity and population growth rate between the period 1976 and 2010. Partial adjustment and ARIMA models were used to provide electricity demand estimation and forecast, and resul ts were compared with government projections. In the partial adjustment model, we found that income and population growth are the main determinants of electricity demand in Ethiopia, while the effect of electricity price is insignificant. Moreover, elasticities of electricity demand in Ethiopia are low (inelastic demand) meaning that, the responsiveness of consumers' to price, income and population growth rate changes is limited. From the comparison of ARIMA (1, 2, 1) model forecasts and government projection, except reference scenario, we found that the government electricity demand projections are consistently higher for moderate and targeted scenario than forecasted values of ARIMA (1, 2, 1) model. Key words: Partial adjustment model, Elasticity, Co-integration Analysis, ARIMA
Analysis of Factors Associated With Car Collisions Resulting in Property Damage in Traffic Accidents (The Case of Addis Ababa
(Addis Abeba university, 2008-12) Hailemariam, Kasahun; Gotu, Butte(PhD)
The numbers of vehicles crashed due to traffic accidents have a great impact on the economy of the societies. There are several approaches that researchers have employed to study this problem. This study will attempt to address factors that affect the number of vehicles crashed resulting in property damage due to road traffic accidents in Addis Ababa. This study is based on a secondary data obtained from Addis Ababa Traffic Control and Investigation Department in 18 month period. This paper is focused on traffic accidents resulted in property damage only. For analyzing this data we use Gamma regression model because this is count data and also under dispersed. In order to establish the required relationships we use 22 explanatory variables with a total of 132 levels. Among the 22 explanatory variables only five variables with a total of 36 levels that have significant association with response variable at 10%  level significance remains in the model for analysis. Findings of this study have shown all levels of light condition, all types of vehicles, all vehicle services years expect vehicles with service years between 2-5 years, all types of collisions expect collision with non-living objects, and all types of road junctions have significant effect on the number of vehicles crashed that resulted in property damage. The result also shows that the coefficients of all vehicle types have negative sign. This indicates that type of vehicles change the mean the response variables; i.e. the number of vehicles crashed per incidence of accidents changes by a value lies between 0 and 1. However, coefficients of levels other variables have positive sign and hence change the mean of the response variables in more than 1. Since the coefficients of level vehicles with service years between 2-5 years and collision with non-living objects are not significantly different from zero, they have no effects on the mean of the response variable
Analysis of Infant Mortality in the Vicinity of Jimma Zone
(Addis Abeba university, 2000-06) Bekele, Fetene; KorMo, Tadewos(PhD)
111ere has been an explosion of interest in fhe (oialysi,;' o{s/o·j,(vl!l'dCa'a in'the' Ihl"t 35 );e{o\, which is resulting in the development of many nell' theoretical ideas (md usejitl methods, specially in the study of the relationship between survival times and explonat01:l' variables, 111C application of survival data analysis is common in different fields ranging fj-om Economics to Engineering. The value of survival analysis techniques is wide in medical statistics. The focus o.f this thesis is on the application of survival data analysis to the data generatedfj-om epidemiological study conducted/i-om 1992 to 1994 in Jimll1a, Keffct-Sheka and Iliubabour zones, J\lethodological procedures to handle problell1s originated fj-OIl1 surviml data are given briefly. A review on determinal1fS o.f hlfcll1t survival/i-om related studies is made. The main objectives o.f the thesis are to determine the survival pattel'l1 up to first birthday and investigate the possible risk factors that contribute most to the early mortality. Demographic, socioeconomic, Environmental status, health service usage and traditional practice indicator variables are considered. The basic Cox regression model is/iffed to get the effect of a variable, a({justedfor other variables, Results are given both in numerical estimates and graphical presentations where applicable. The descriptive analysis shows that neonatal mortality rate is 26.6 pel' 1000 live births, postnatal mortality rate is 73,1 per 1000 live births, and the overall 111fant Mortality Rate (IMR) is 101,9 pel' 1000 live births. The cumulative survival probabilities are 11 ().9855(.\·.e~O.{)013). 0.9736(s.e~0.0018), O. 9039(.1. e. =0.(033) for days 7, N! and 36{), re.~peClively. The final Cox's regression model sholl's that breastIeeding practice, \'{{ccination status at birth. weight at birth. .klll/ily size, death of previous children, sex of {[Ii iJ?fant, maternal age, marital status, visit to maternal clinic, sll'allowing blllter, place of deliveJ}" and availability of latrine facility are significaJ1lly important variables in determining iJifaJ7fs' chance of survival. Analysis by age of iJ?fams sholl'S that socioeconomic and environmental variables are iJif/uential at later ages of life. Also analysis by place of residence sholl'S that the effect -of these variables is important only at urban areas. Finally, relevant discussion is made and possible recommendations are given according~l' to interested policy makers and/i'ont-line health workers.
Analysis of Risk Factors for Under-Five Child Malnutrition in Ethiopia: Multi-Level Approach
(Addis Abeba university, 2016-06) Workineh, Sintayehu; Teshome, Birhanu(PhD)
Although the problem of malnutrition affects the entire population, children are more vulnerable because it reduces their physical growth, proper organ formulation and cognitive development and weakens immune system. Under-nutrition contributes half of all deaths in children underfive in Africa and Asia. The solution of this big public health problem is essential. From this perspective, this study aimed to explore risk factors for under-five child malnutrition in Ethiopia. The data for the study were taken from Ethiopian Mini Demographic and Health Survey (EMDHS) of year 2014. To achieve objectives of the study, two indicators (stunting and wasting) were studied. The two indicators were treated separately due to their biological differences. Statistical models that handle the complexities of correlated data were employed. Generalized Estimating Equations (GEEs), Alternating Logistic Regression (ALR), Proportional Odds Model (POM), Partial Proportional Odds Model (PPOM) and baseline category logit models were used for analysis. Results showed that mothers with less educational level, preceding birth interval less than 24 months, from Tigray, Affar and SNNP regions were associated with higher probability of stunting. Child age in months had positive effect on child stunting. Children from non-educated mothers, from Affar and Somali regions had higher chance of wasting. Two-level ALR analysis indicated strong association ( ) between two children from the same households. In conclusion, results suggest that region, mother’s education level, wealth index, preceding birth interval and child age were associated with stunting, while region, mother’s education level, mother’s current marital status, sex and age of child in months were associated with wasting among children under-five in Ethiopia. To reduce under-five child malnutrition, some crucial steps regarding educating mothers and improving economic situation of population as well as supplementary feeding programs should be considered. Key Words: Alternating Logistic Regression (ALR), Generalized Estimating Equations (GEEs), Proportional Odds Model (POM), Partial Proportional Odds Model (PPOM) Baseline Category Logit, Stunting, Wasting
Analysis of Stem Borers and their Parasitoids in Maize and Sorghum Agro-ecosystems in Eastern Ethiopia using Generalized linear Model
(Addis Ababa University, 2008-06) Gemechu, Bedanie; Taye, Girma (PhD)
Maize and Sorghum are the most important food crops in Ethiopia. However, stem borers became the major problem resulting in yield losses. A study was conducted in the cenh'al Rift valley of Ethiopia to find out stem borers infestation, diversity and their parasitoids interaction, to determine the appropriate statistical model that should be used in the analysis of this sort of data. The analysis of stem borers and their parasitoids in maize and sorghum agro-ecosystems was made by generalized linear model using data recorded from eastern Ethiopia in 1999 and 2000. In previous studies, some researchers have used general linear Model (GLM) analysis on the stem borers' data. But, general linear model is used for data satisfying assumption of normality where as there are situations in which non-normal data is treated. This includes using generalized linear model analysis that allows the use of different exponential distributions and non-linear link functions. Furthermore, generalized linear model was found to be relatively better than general linear model for the analysis of data which violate the assumption of linearity and normality. Moreover, the response variables: infestation and diversity of stem borers were found to be discrete and counts. The analysis of the data using this model and appropriate link functions is applied to identify the significant predictors which contribute positively or negatively to the diversity and infestation of stem borers among the explanatory variables. Accordingly, generalized linear model was found to be the best model in identifying the significant predictors. Consequently, infestation and diversity of stem borers were not significantly affected by the season, wild host and the predator species. It was also observed that the infestation was not affected by cropping systems where as the diversity is not affected by year, pest species and crop growth stages. Mean separation was made by using LSD and Duncan's multiple range tests to see the differences with in the explanatory variables. There fore, the effects of vegetation types across locations on the infestation and diversity of stem borers is not significantly different from district to districts at 95% confidence. Besides, there is no significant difference ofinfestation of stem borers with or with out the wild hosts where as diversity does not differ significantly for the main crops (sorghum or maize), cropping systems (mixed or sale), for the parasitoid species (presence or absence) and for the nitrogen contents. But, there is significant difference in infestation of stem borers between the years, 1999 and 2000, between the main crops, more for sorghum (47.812) than maize (19.775), with in the pests species which high for the key species (44.867) than other species (18.639), with in the growth stages of the crops in which more infestation was measure a during the stubble stage (55.00) followed by maturity stage (43.463) the vegetative stage (29.683) at 95% level of significance. The presence of the parasitoids species resulted in more infestation than its absence. The interaction effects of year and location basically implies the difference of infestation and diversity of stem borers in the years 1999 and 2000 across the locations. This is due to the fact that different locations have specific determinant factors such as temperature, altitude and rain fall and etc. which are not included in this study. Finally, it is recommended that generalized linear model is flexible, easy to use for any type of data, due attention should be given to the highly infested areas, control should be devised for the stem borers and great care should be given for the parasitoids during the use of chemicals.
Analysis of the Determinants of Ethiopian Trade Balance: an Ardl Approach
(Addis Abeba university, 2016-06) Tagele, Molla; Gottu, Butte(PhD)
Variables that determine the situation of a given economy could have short or long-run relationships. This study investigates the short and long-run relationships between the trade balance, real gross domestic product, real effective exchange rate, and inflation rate in the case of Ethiopian economy. The bounds testing approach to co-integration and error-correction models, developed within an autoregressive distributed lag(ARDL) framework is applied to quarterly data for the period 1993 to 2015in order to investigate whether a long-run equilibrium relationship existsbetween the trade balance and the variables indicated above. The result indicates that in the long-run one per cent increase in real effective exchange rate increases trade balance by about 0.19 percent. Additionally, variance decompositions (VDCs) and impulse response functions (IRFs) are used to draw further inferences. Under Johansen co-integration approach the error correction model (ECM) shows both the short- run and long -run relationship among the indicated macroeconomic variables (real effective exchange rate, real gross domestic product and inflation).The coefficient of the error correction term indicates that 75.7% of shock (short-run disequilibria) will be adjusted within the same year, but under bound test of co-integration nearly 80 % of the previous quarter’s shock adjusts back to long-run equilibrium in the current quarter
Application of Balanced Incomplete Block Design in ps Sampling Without Replacement
(Addis Abeba university, 2007-07) Tefera, Yigrem; Sharma, M.K. (Professer)
In this study a procedure for selecting a sample of size n with probability proportional to size without replacement was proposed. This procedure used the combinatorial properties of balanced incomplete block design. An application of this procedure has been illustrated by using secondary data, obtained from National Labour Survey conducted in 2005 by Central Statistical Agency of Ethiopia. Further, the variance of the estimated total unemployed population was compared by using Horvitz-Thompson (1952) and Sen –Yates- Grundy (1953) estimators. The result suggests that, variance due to Sen-Yates –Grundy is better estimator than Horvitz-Thompson . In addition, I would like forward that up to date investigation in the work of application of balanced incomplete block design in pps sampling without replacement at national and/or regional level is important in minimizing cost at large. Researchers who try to examine the application of BIBD in pps sampling without replacement in the further have to take into account those variables other than unemployed population
Application of Cox Proportional Hazards Model in Case of Tuberculosis Patients in Selected Addis Ababa Health Centers, Ethiopia
(Addis Ababa University, 2013-06) Tolosie, Kabtamu; Sharma, M.K. (Prof.)
Tuberculosis (TB) is a chronic infectious disease and mainly caused by mycobacterium tuberculosis (MTB). It has been one of the major causes of mortality in Ethiopia. The objective of the study was to identify factors that affecting the survi val of the patients with tuberculosis who started treatment for tu berculosis. This was a retrospective study in six randomly se lected health centers in Addis Ababa, Ethiopi a. The data were obtained from med ical records ofTB pat ients regi stered from September 2012 to August 20 13 and treated under DOTS strategy. Out of the total 826 registered TB patients 105 ( 12.7 1%) di ed during the study period and 7 12(87.29%) were censored . Based on Kaplan Meier survival curves, log rank lest and Wi lcoxon tcst it was found that the patients had statistically significant d ifferences in survival experience with respect to age, body weight at initiation of treatment, TB patient category and HIV status. Multivariable Cox hazards regression anal ysis revea led that the covariates age, TB patient category, HIV and age by HIV interaction were significant risk factors associated with death status in TB patients.
The Application of D2 Statistic to Measure Genetic Divergence in Different Bread Wheat Genotypes
(Addis Abeba university, 2014-06) Mekonnen, Leul; Sharma, M.K. (Professer)
Wheat is one of the most important cereal crop in Ethiopia, ranking third in total production next to maize and teff. Wheat covers a total arable land of 110,434 ha with average productivity of about 8.4 qt /ha, which is below the national production average (14.4 qt /ha). The objective of this research was to study the genetic dissimilarity among twenty bread wheat genotypes. The study is made based on twenty bread wheat genotypes which was obtained in 2007 and 2008 from trials conducted by the Ethiopian Institute of Agricultural Research (EIAR) and were evaluated for their diversity to estimate the genetic divergence and clustering them into homoGenous groups for the hybridization program in 8 environments using Mahalanobis D2 statistic. The D2 value data can be used in cluster analysis to identify groups of related cultivars. Using UPGMA clustering technique all the Geno types were grouped into six cluster on the basis of D2 value using Tocher's optimization technique. The intra- and inter-cluster D2 values suggested that within cluster genetic diversity is narrow, but the genetic diversity among clusters is greater. Therefore, The presence of significant genetic variability among the evaluated bread wheat genotypes suggests an opportunity for improvement of grain yield through hybridization of genotypes from different clusters and subsequent selection from the segregating generations would broaden the genetic base of bread wheat breeding populations. Thus, those analysis based on multivariate methods using D2 statistic is useful in providing information for more efficient variety development programmes.
Application of Garch Models in Forecasting the Volatility of Export Prices
(Addis Abeba university, 2009-02) Terefe Amare; Abegaz Fentaw (PhD)
This thesis discusses the GARCH model fitting and volatility forecasting of export prices data. We have chosen to confine our analysis on total export prices, coffee export prices and oil seeds export prices. The implied goal is to fit an appropriate GARCH model, to find out if the mentioned export prices are volatile or not and to forecast the volatility for some future times. ARMA(1,1) is found as the most appropriate model for the conditional mean of total, coffee and oilseeds export prices and it is also found that GARCH(2,1) for modeling volatility of total export prices and coffee export prices, and GARCH(2,2) for modeling volatility of oil seeds export prices as best models. Moreover, the results suggest that the export prices volatility is persistence in all the three cases indicating that past volatility is important in predicting (forecasting) future volatility. Key Words: export, GARCH, volatility, forecasting, Ethiopia
Application of Linear Mixed Model to Incomplete Block Designs
(Addis Ababa University, 2010-06) Lakew, Demeke; Taye, Girma (PhD)
The study was designed to examine th e application of linear mixed model to incomplete block designs. In addition to thi s, it is planned to compare VARCOMP, ML and REML estimation methods for variance components of linear mixed models. Sixty three promising bar ley lines and one standard check cultivar which were obtained from EIAR have been evaluated for grain yield performance and adaptation across eight environments (combination offour lo cations by two ferti lizers). The mean grain yields for individual line ranged from 17.06 to 33.21 quintals per hectare and the mean grain yields for in dividual environment ranged from 16.80 to 44.214 quintals per hectare. The highest mean grain yield was observed at BEKOjl, while the lowest mean grain yield was registe red at SHENO with both fertilizers doses (100 and 150 kg). When we compare each variety with specific environment, lin e 8 and 55, Variety 20 and 14, Variety 49 and 54 and Variety 48 and the Local check to be ada pted and have best mean effects to BEKOJI, SHENO, HOLETTA and NORTH GONDAR respectively. KEML estimates for vari a nce components are in distinguishable from classical techniques in case of balanced data. This im pl ies optimal minimum variance properties and REML estimates do not rely on normality assumption. But, for unbalanced data, the REML estimation for variance components is different from classical estimates. We have seen from the diffe rences that estimation of variance components benefits from REML bu t VARCOMP does not. Therefore, the REML approach is appropriate to estimate va riance components in SLD and SLD with missing plots. Varieties/lines listed above were recommended to release in Ethiopia that have similar ecologic zones of respective location s and REML techniques was recommended to be used for va riance components esti mates of linear mixed model for SLD and SLD with missing plots. It was found that diagnostics for Linear Mixed Model applied to estim ate variance components are perh aps an area that needs exploration in the future.
Application of Longitudinal Count Data Models to Progression of CD4 Count: A Case of Debre Markos Referral Hospital
(Addis Abeba university, 2017-06) Desyebelew, Belay; Teshome, Birhanu(PhD)
Even though the world is ghting HIV disease in unity and patients are getting antiretroviral therapy treatment, HIV disease continues to be a serious health issue for parts of the world and large number of AIDS related deaths are being registered every year. A number of studies have been conducted to assess factors related with the progression of the disease using surrogate end- points like CD4 cell count. The main objective of this study was to make use of appropriate statistical models to analyze CD4 cell counts data and identify associated risk factors a ecting the CD4 cell progression of patients under ART tra etment in Debre Markos Re eral Hospital. In this longitudinal retrospective cohort based study, data was collected from 445 HIV patients reg- istered for ART treatment between September, 2005 and August, 2014 in the Hospital. Poisson, Poisson-Gamma, Poisson-Normal, and Poisson-Gamma-Normal models were applied to account for overdispersion and correlation in the data. Poisson-Gamma-Normal model with random in- tercept was selected as a best model to t the data based on di erent model selection criteria. The ndings of the study revealed that time in months, sex of patients, baseline WHO stage and baseline CD4 cell count were found to be signi cant factors for progression of HIV patients' CD4 cell count. Patients who started ART at higher baseline CD4 counts evolved higher than those who started at lower CD4 counts. Therefore, patients should start ART treatment early to increase their CD4 cell count progression. Keywords: CD4 count, Longitudinal data analysis, Poisson-Normal Model, Poisson-Gamma- Normal model, Antiretroviral therapy (ART), HIV/AIDS
Application of Mixed Model in Agricultural Field Experiment on Wheat
(Addis Abeba university, 2008-01) Kedir, Abbas; Taye, Girma
This study was designed to explore the efficiency of mixed model over fixed effect model. Furthermore, it was designed to determine how the REML procedure was used to find the estimate of variance-covariance matrix of the model. The methods of restricted maximum likelihood are applied to the data from uniformity and multi-location trials. In this study, two data sets are used; uniformity trial which was used to determine optimum plot size and shape, and multi-location trial which is used to identify the effect of location on different varieties of wheat. In uniformity trial, two sowing methods: row and broadcast sowing, are considered both in 1996 and 1997 in which the trial were conducted. These factors are considered as fixed effect of the model. But, plot sizes ranging between 1 and 24 m2 with different shape are taken as random factors of the model. Regarding Multi-location trial, it was conducted in 2001, 2002 and 2003 using 20 wheat varieties in 14 locations that were expected to have the same agro-ecological effect for wheat adaptations. All the three years in which the trail was conducted are taken as fixed effect whereas varieties and locations are considered as random factors of the model. The result of the study based on restricted maximum likelihood (REML) revealed that small plot sizes have larger coefficient of variation in both 1996 and 1997 for both row and broadcast sown methods. This indicates that the coefficient of variation and plot sizes are inversely proportional. Nevertheless, the coefficients of variation for 1996 sown by broadcast method have higher values than the other year of the same method. This shows that the coefficient of variation was influenced not only by soil variability and plot orientation but also by crop geometry resulting from the sowing method. The test done for identification and adaptation of different varieties on different location shows that there is no average effect due to variety and hence all varieties have the same effect on the response variable. This implies that all varieties are contributing the same average effects for the yield obtained. 9 However, the test statistics for location shows that the coefficients for Bako, Bekoji, Debre Zeit, Ginchi, Hollota, Kulumsa, and Sinana are significantly different from the remaining locations. This significance indicates that the average productivity potential of these locations is higher than that of the remaining locations. In general, the contribution of each location to yield is not the same unlike that of variety. Hence, the average effect of location on the response variable is not the same.
Application of Spatial Mixed Model in Agricultural Field Experiment
(Addis Ababa University, 2010-06) Bayisa, Dibaba; Taye, Guma (PhD)
The objective of this study was to evaluate the efficiency of spatial stati stical ana lysis in field trials and, particularly, we have demonstrated the benefits of the approach when experimental observations are spatia lly dependent. We have compared (i) cl assical randomi zed complete block model with independent and identica lly distributed errors (RCBDiid); (ii) the most common spatial models: Exponentia l, Spherical and Gaussian (with and without nugget effect); (i ii) spatial models of (ii) with and without block effects; (iv) complete random model (CR). The data used in this study were obtained from separate trials for Bread wheat and Durum wheat wh ich was coded as BW99RVTII (Regional Varity Trial Two of Bread Wheat conducted in the year 1999 Eth.C.) and DW99RVTII (Regional Varity Trial Two of Durum Wheat conducted in the year 1999 Eth.C.), respectively. The restricted maximum likelihood (REML) method was used for est imation of spatial covariance parameters. The denominator degree of freedom of F test for treatment effects were computed using the Kenward-Roger method for a ll models (Kenward and Roger, 1997). Semivar iogram were used for the initial estimate of the parameters of the spatial covari ance structure. Akaike's Information criterion (AIC) and Corrected Akaike's Information cri terion AICc and the Likelihood ratio test were used for model comparison. The result showed that in all of the four trial s, the estimated residual from RCBiid model were s ignifi cantly spat ially correlated, providing evidence of field heterogeneity within blocks that cou ld make the RCBiid method less powerful than a method that incorporated the spati al correlation. Furthermore, the finally se lected spatial model for a ll data set showed that blocking seemed to be unnecessary if these se lected spatial model were used for each of the trial s. The results also showed that the spatial models provided a smaller p-value and standard error than the class ical analysis of variance models. We have concluded that randomizat ion and blocking does not completely remove spatial variation. Hence, we recommended that researchers should take into account spatial variation in field experiment in s ituations were spat ial dependence among the observations is sign ificant. Key words: Spatial Models; Semivari ogram
Application of Spatial Modeling to the Study of Soil Fertility Pattern
(Addis Abeba university, 2008-09) Obsi, Dechassa; Taye, Girma (PhD)
Spatial statistical analysis was undertaken to study the variability of potato yield data due to soil fertility pattern. Seed potato yield were measured from field uniformity trial conducted at Hollota and Kulumsa agricultural research centers in year 2001 on an area of 0.15 hectare at each site. The harvested area was divided in to basic units of 1.2x1.5 m and a total of 1658 and 931 plots were considered from Hollota and Kulumsa respectively. The basic units were combined in different plot sizes to acquire the required plot dimension. In this study, the Monte Carlo test for completely spatial randomness is applied and the result shows no complete spatial randomness detected in the series of potato yield data for both sites. Thus, to set a model adjusted for spatial pattern, Moran’s index and Geary’s coefficient were applied to test for global and local spatial autocorrelation respectively. The result shows posit ive spatial autocorrelation detected among potato yield data using Rook’s weighted neighboring plot relations. The result also shows, increasing plot size will not generally make the observed spatial autocorrelation insignificant. An autoregressive model, that is adjusted for presence of spatial autocorrelation in simulated plot size of 12m2 is fitted for row and column effect for each site. The result shows significant positive association between neighboring plots row effect and the adjusted potato yield. In addition, the result from comparison of model adjusted for presence of spatial autocorrelation and conventional OLS based analysis of variance shows the autocorrelation parameter accounts significant percent of variation among potato yield in both sites

Browse

Browsing Statistics by Title

Results Per Page

Sort Options