dc.contributor.author | Koroba, Tom J | |
dc.date.accessioned | 2013-03-14T08:16:33Z | |
dc.date.issued | 2010 | |
dc.identifier.citation | Master of Science in Information Systems | en |
dc.identifier.uri | http://erepository.uonbi.ac.ke:8080/xmlui/handle/123456789/13718 | |
dc.description.abstract | The purpose of this research is to learn CPU and memory utilization data to derive
more information from the data, that would help information technology
organizations like telecoms mitigate some problems and offer opportunities for
capacity planning and cost optimization for CPU and memory resource requirements
for information technology systems.
Classification problems for various data set concepts have been attracting a significant
amount of research attention using machine learning algorithms. Raw data with
attribute-value or instance-value has been the most frequently encountered data type
which can be presented on a Euclidean space. Classification methods are used for
discovery, recognition, grouping and prediction
The CRISP-DM methodology was used for this research and various ML algorithms
for data classification and prediction were reviewed and accessed. This study showed
that supervised ML and in particular KNN was appropriate for the research problem.
For more meaningful information and decision making, extra features on the data
were required to determine resource utilization levels. Using the Weka software suite,
the study showed that there were interesting prediction results which were obtained by
learning the data to predict required memory and CPU loads given that some level of
optimization was required.
Prediction results for CPU and memory resources are discussed as well as those from
tree type J48 and SVM methods are compared. Finally, recommendations and
conclusion are given. | en |
dc.description.sponsorship | University of Nairobi | en |
dc.language.iso | en | en |
dc.publisher | University of Nairobi | en |
dc.subject | Information systems | en |
dc.subject | Resource capacity planning | en |
dc.subject | Machine learning algorithms | en |
dc.title | Information systems resource capacity planning using machine learning algorithms | en |
dc.type | Thesis | en |
local.publisher | School of Computing and Informatics | en |