Trends in Data Mining and Knowledge Discovery - Advanced Techniques in Knowledge Discovery and Data Mining - page 9

or useful. In this case, we may choose different DM tools and repeat Step 4 to
extract new and potentially novel, interesting, and thus useful knowledge.

The feedback paths are shown as dashed lines in Fig. 1.3. It should be
understood that the described feedback paths are by no means exhaustive.
Table 1.1 compares the six-step DMKD process model with the nine-step
DMKD process model [31] and the five-step model [16].
The common steps for the three models are domain understanding, data
mining, and evaluation of the discovered knowledge. The nine-step (Fayyad's)
model is very detailed, and although it provides the most guidance, it performs
Steps 5 and 6 too late in the process. We think that these steps should be
performed during the steps of understanding the domain and understanding the
data, to guide the process of data preparation. In other words, the goal of data
preparation is to prepare the data to be used with the already chosen DM tools,
while their model suggests that the DM tool is selected in Step 6, depending on the
outcome of data preparation. This may cause problems when choosing a DM tool
because the prepared data may not be suitable for the given tool. Thus, an
unnecessary feedback loop may be needed to change data preparation in Steps 2, 3,
and 4. The five-step (Cabena's) model is very similar to the six-step (Cios's)
model, except that the data understanding step is missing. The incompleteness of
the Cabena model was pointed out in [39] where the author used it in a business
domain, and one of the conclusions was the necessity of adding one more step
(forbreaking…)

Table 1.1. Comparison of three DMKD process models.

6 Step DMKD Process
[18]

9 Step DMKD Process
[31]

5 Step DMKD Process
[16]

1. Understanding the
domain

1. Understanding application
domain, identifying the
DMKD goals

1. Business objective
determination

2. Understanding the data 2. Creating target data set

3. Data cleaning and
preprocessing
4. Data reduction and
projection
5. Matching goal to
particular data mining
method

2. Data preparation

3. Preparation of the data

6. Exploratory analysis,
model and hypothesis
selection

4. Data mining 7. Data mining 3. Data mining

5. Evaluation of the
discovered knowledge

8. Interpreting mined
patterns

4. Analysis of results

6. Using the discovered
knowledge

9. Consolidating discovered
knowledge

5. Knowledge
assimilation