Phase 1: (3 marks)
You should perform a literature review on the above topic and
include all relevant references in your report, no more than 2,000 words.
You should cover the topic in a coherent manner. This must be done by reference to the
relevant literature, so the bibliography is an important part of the research
and should contain complete references for current or recent research in the
field. You should abstract information from the literature you read using your
own words, instead of merely repeating what is said on the sources you
use. Cite all relevant references,
including on-line references. Failure to
do this could result in a penalty due to plagiarism. Your research must be
your own work. Check your final document using SafeAssign Tab in Blackboard.
You will need to search in a digital library to do this
research. You can find the necessary information in scholarly sources, which
reside on the SDL (Saudi Digital Library) using Microsoft Edge.
Note: Science and technology rapidly advances;
therefore, "old "stuff," other than as background information,
can be misleading and lead to wrong conclusions. Look for possible topics and
background information in specialized encyclopedias, we preferred recent
resources from 2015 and after, but if you need and old one for background
information you can back to these resources to clarify concepts.
Phase 2: (3 marks)
Use WEKA to analyze the dataset you choose with different classification
algorithms similar to the work did in the attached paper:
what the researcher did in his research (scientific paper attached) to obtain
similar results (Experimental and results section).
your result and present it in clear way.
your results with the researcher's results.
any difference between the results.
Phase 3: (2 marks)
to improve the researcher results by applying the same dataset in to different algorithm
(classification methods, pre-processing steps, parameters) then, observing
their impact on the results.
1. In the decision
tree algorithm J48, you can use different values of the confidence factor
parameter which determines the amount of pruning.
2. In KNN classifier,
you can vary the value the parameter K.
3. Use different
feature selection method as pre-processing step to choose a subset of
attributes or construct new attributes and see the impact on the results.
When you deliver any improvement in the results,
Show the process and the results in details:
setting did you try?
reliable are the results?
evaluation measure did you use to compare your results with the researcher
Copyright © 2021 | Truelancer.com