Designing and Building a Data Mining Model with Decision...

Preview:

Citation preview

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Designing and Building a Data Mining Model with Decision Tree

Algorithm

1) Creating Data Source View :

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

2) Creating and deploying Mining Structure :

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

· Default Algorithm Parameters :

COMPLEXITY_PENALTY: 0

MINIMUM_SUPPORT: 10

SPLIT_METHOD: 3

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Here we got bellow decision tree.

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

· Update Algorithm Parameters :

COMPLEXITY_PENALTY: 0.5

MINIMUM_SUPPORT: 10

SPLIT_METHOD: 2

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Compare to default tree, here total levels of the tree were reduced to 8 levels.

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

· Update Algorithm Parameters :

COMPLEXITY_PENALTY: 0.9

MINIMUM_SUPPORT: 10

SPLIT_METHOD: 2

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Compare to default tree, here total levels of the tree were reduced to 5 levels but not able to

predict properly up to last level.

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

· Update Algorithm Parameters :

COMPLEXITY_PENALTY: 0.5

MINIMUM_SUPPORT: 50

SPLIT_METHOD: 2

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Compare to default tree, here total levels of the tree were reduced to 6 levels. And ability for

prediction of the bike buyers were good.

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

· Update Algorithm Parameters :

COMPLEXITY_PENALTY: 0.9

MINIMUM_SUPPORT: 50

SPLIT_METHOD: 2

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Compare to default tree, here the levels of the tree were reduce to 6 level.

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

· Update Algorithm Parameters :

COMPLEXITY_PENALTY: 0.9

MINIMUM_SUPPORT: 100

SPLIT_METHOD: 2

Cleveland State University Dhruv R Patel Lab Assignment 4 CIS 660

CSU ID : 2652790

Compare to default tree, here total levels of the tree were reduced to 4 levels but not able to

predict properly. Here because of the large minimum support some important data for the

decision were missing.

3) Conclusion: By comparing all above trees of different parameters, below parameters

were best as the MINIMUM_SUPPORT was not too small and too large, so it not

exclude high priority data and not include low priority data.

COMPLEXITY_PENALTY: 0.5

MINIMUM_SUPPORT: 50

SPLIT_METHOD: 2