Database Reference
In-Depth Information
and collaborative teamwork. The distribution of percentage marks under the stage
headings in Table 1 is an example scheme practised by this author.
According to the framework, a project can be broadly classified into one of the fol-
lowing four categories:
1.
Unsatisfactory. This kind of projects normally has major flaws in project activities
at certain stage(s). Without a clear discovery aim, random decisions are made and
random actions are taken. Data are not examined and well prepared before mining.
Certain solutions with default parameter settings are chosen without any good
reasons. Irrelevant patterns with little interpretation are collected as result. Little
attention is paid to evaluation of the result patterns. The poor quality of work re-
flects no serious attempt. The total mark awarded for this category should be
lower than the bare pass mark (e.g.40%).
2.
Fair. This kind of projects normally produces some positive results in terms of
experience. However, the work is not well planned. It involves either too many
trial-and-error or limited project activities. Students show limited understanding of
knowledge and make decisions without full consideration of the issues concerned.
Some directly “copy-and-paste” style references to seen examples are made with-
out questioning the relevance. Limited understanding about evaluation of result
patterns is evident. The total percentage mark for this category is between the pass
and a lower 2.II (e.g. between 40% and 54%).
3.
Good. This kind of projects shows sufficient understanding and good application
of knowledge. The objectives are clear. Project activities are planned and targeted
not randomly decided. Actions taken in preparing data are well thought with good
supporting arguments. The selection of mining methods is sensible, and so are set-
tings of relevant parameters. Alternative methods or alternative parameter settings
are tried with justification. There are clear evidences of appreciation of evaluation
of result patterns. Sensible interpretations and implications for course of actions
are drawn from the mining results. The total percentage mark for this category is
between a higher 2.II and a high 2.I (e.g. 55% and 69%).
4.
Excellent. This kind of projects shows all the merits of the category above and fur-
thermore demonstrates excellent performance throughout the entire process. A sense
of critical analysis and critical evaluation is demonstrated at every stage. There is a
well-thought reasoning from a business objective to data mining tasks. All decisions
and actions in data preparation and pre-processing are supported by sensible argu-
ments. Data mining tasks are well defined and relevant to the aim. Each trial of data
mining serves a clear purpose. The evaluation of resulting patterns is thorough and
appropriate, and influences the selection of useful patterns for potential use. The to-
tal percentage mark for this category is a clear first (e.g.
≥
70%).
In practice, the marking can be done in a top-down or a bottom-up fashion. In the
bottom-up approach, detailed percentage marks are first given to each stage of the
project according to the factors outlined in the assessment framework. The total mark
for the whole project should then reflect the appropriate category for the project. In
the top-down approach, the assessor first classifies the project into one of the catego-
ries according to the category descriptors, and then fine tune the percentage marks for
each stage to reflect the description of the category.
