Chapter 3 : Start to read

I am not a good reader. Whenever I stay at the library will end like these :

Get books

Get to read

Get bored

Stress

Get hungry

But I decide to change the habit. I will learn to be a good reader every now and then. Start to read with these three papers :

• Association Rule Mining as a Data Mining Technique -> Irina Tudor
• Mining Association Rules with Apriori -> Jinbo Paul Lin's resume
• Application using Data Mining Association Rules with Priori Method for Analysis of Data on The Market Basket Pharmacy Sales Transaction -> Leni Meiwati

Paper 1 :

• Data Mining devides into two major classes which are Supervised ( Bayesian, Neural Network, Decision Tree, Genetic Algorithm, Fuzzy Set, K-Nearest Neighbor ) and Unsupervised ( Association Rules and Clustering )

• Association rules :

    a. Support
    b. Confidence

Typically, association rules are considered interesting if they satisfy both a minimum support threeshold and a  minimum confidence threshold.

Case study : Market Basket Analysis

Association rule mining searches for interesting relationships among items in a given data set. Considering the example of a store that sells DVDs, Videos, CDs, Books and Games, the store owner might want to discover which of these items customers are likely to buy together.

• Suppose minimum support required is 2
• Let minimum confidence required is 60%
• We have to find out the frequent itemset using Apriori algorithm
• Association rule will be generated using the two parameters minimum support and minimum confidence

 Transaction :

1. Customer A bought BOOKS, CD, VIDEO
2. Customer B bought CD, GAMES
3. Customer C bought CD, DVD
4. Customer D bought BOOKS, CD, GAMES
5. Customer E bought BOOKS, DVD
6. Customer F bought CD, DVD
7. Customer G bought BOOKS, DVD
8. Customer H bought BOOKS, CD, DVD, VIDEO
9. Customer I bought BOOKS, CD, DVD

• JOIN STEP

1- Itemset

*/ { BOOKS }, support count 6

*/ { CD }, support count 7

*/ { VIDEO }, support count 2

*/ { GAMES }, support count 2

*/ { DVD }, support count 6

2-Itemset

*/ { BOOKS, CD }, support count 4

*/ { BOOKS, VIDEO }, support count 2

*/ { BOOKS, GAMES }, support count 1

*/ { BOOKS, DVD }, support count 4

*/ { CD, VIDEO }, support count 2

*/ { CD, GAMES }, support count 2

*/ { CD, DVD }, support count 4

*/ { VIDEO, GAMES }, support count 0

*/ { VIDEO, DVD }, support count 1

*/ { GAMES, DVD }, support count 0

The red are called prune step. Any item that has a support count less than the minimum support count required is removed the pool of candidate items.

3-Itemset

*/ { BOOKS, CD, VIDEO }, support count 2

*/ { BOOKS, CD, DVD }, support count 2

4-Itemset

*/ { BOOKS, CD, VIDEO, DVD }

• FINAL STEP

The final step is to provide the association rules from frequent itemsets.

• For each frequent itemset " a ", generate all none empty subset of " a "
• For every nonempty subset " s " of " a ", output rule " s -> ( a-s )" if support count (a) / support count (s) >= min_conf

For example L = { BOOKS, CD, VIDEO }. Its all none empty subsets are { BOOKS, VIDEO }, { BOOKS, CD }, { CD, VIDEO }, { BOOKS }, { VIDEO }, { CD }

Let minimum confidence threshold is 60 %


R1 : BOOKS and VIDEO -> CD
     Confidence = support count { BOOKS, CD, VIDEO } / 
                  support count { BOOKS, VIDEO }
                = 2/2
                = 100 % - R1 is selected


R2 : BOOKS and CD -> VIDEO
     Confidence = support count { BOOKS, CD, VIDEO } /
                  support count { BOOKS, CD }
                = 2/4
                = 50 % - R2 is rejected


R3 : CD and VIDEO -> BOOKS
     Confidence = support count { BOOKS, CD, VIDEO } /
                  support count { VIDEO, CD }
                = 2/2
                = 100 % - R3 is selected


R4 : BOOKS -> VIDEO and CD
     Confidence = support count { BOOKS, CD, VIDEO } /
                  support count { BOOKS }
                = 2/6
                = 33 % - R4 is rejected


R5 : VIDEO -> BOOKS and CD
     Confidence = support count { BOOKS, CD, VIDEO } /
                  support count { VIDEO }
                = 2/2
                = 100 % - R5 is rejected


R6 : CD -> BOOKS and VIDEO
     Confidence = support count { BOOKS, CD, VIDEO } /
                  support count { CD }
                = 2/7
                = 28 % - R6 is rejected


In this way, we have found three strong association rules.

Paper 2 :

Preparation :

a. JAVA

   * JDK

   * NetBeans

b. MICROSOFT ACCESS

From this paper, I learned how to design system with DFD - Data Flow Diagram :

I also learned how to design structure in database - Microsoft Access :

Paper 3 :

I learned more about JAVA in this paper. I knew pseudocode of the algorithm, the join step and also the prune step.

I also learned how to create Graphical User Interface ( GUI ), classes's structure and another reference of input :

Graphical User Interface

**********

My question is how all of those references help me to finish my final assignment ??

Figure it out now !!!