kmeans
TRANSCRIPT
Hardware And Software Requirements
HardWare:512MB RAM
Software Requirements:Java Run Time Environment
Database –IBM db2
Literature Survey
1. Sequencial Pattern1.1. Priori Based Methods
1.1.1. Apriori All1.1.2. Apriori Some1.1.3. Gsp1.1.4. Spirit 1.1.5.Spade
1.1.5.1. Breadth First Search
1.1.5.2. Depth first search
1.6. Pattern Growth method1.2. Free span1.3. Prefix Span
2. Fuzzy set theory3. Fuzzy Sequencial Patterns
Literature Survey
• Apriori All• Spirit• Spade
Breadth First search Depth First search
• Pattern Growh method• FreeSpan• PrefixSpan
Literature Survey
• Apriori All vs Apriori Some:As AprioriSome is concentrated into
maximal sequences and skipping the smaler sequences which cannot be the solution.
Apriori Algorithm
Apriori Some AlgorithmForward Phase
Apriorisome AlgorithmForward Phase
First generate the first candidate set
For each of the itemset, calculate the fsupp value. Check if the fuzzy support is larger than the threshold then go to forward phase
Else go to backward phase
Forward phase Apply join and prune step
Apriori Some AlgorithmForward Phase
Apriori Some Algorithm Backward Phase
Backward Phase: For all lengths which we skipped:
Delete sequences in candidate set which are contained in some large sequence.
Count remaining candidates and find all sequences with min. support. Also delete large sequences found in forward phase which are non-
maximal.
Apply both these steps untill no more frequent itemset can be generated
Modules
• Finding Fuzzy sequence
• Applying K-means Clustering
• Using Apriori Some Algorithm finding the frequent pattern
Steps for finding the fuzzy sets from the database
• Use the clustering techniques (K-means algorithm).
• Determine the membership function.
Apriori Some AlgorithmPhase 1
Sort The Database
Transforms the QuantitativeValues into fuzzy sets
Find the all large Fuzzy sequence
Apriori Some AlgorithmPhase 2
Forward Phase
Count the Support count
Backward Phase
AprioriSome Algorithm
Less than Threshold
Larger Than
Threshold
Find the Desired Sequence
End
Fuzzy sequential patterns mining with AprioriSome algorithm
1. Analyse the transaction Data.
2. Count each scalar cardinality of each region Rk in the transaction (count jl).
3. For each Pjl, checkwhether countjl is larger than or equal tothe predefined minimum support for each Pjl, j =1–m, l =1–k.
4. If Pjl satisfies this condition, then it is a large 1-fuzzy sequence. L1={Pjl |Count jl>= , j=1-m, l=1-k).
Fuzzy sequential patterns mining with AprioriSome algorithm
5. Then we need to find the large k-fuzzy sequence. Set r =1.
6. Generate the candidate set Cr from Lr=1.
7. Implement the next step until Lr is null, or set r = r - 1 andrepeat Step 5.
8. In this step, we use AprioriSome algorithm to produce thedesired patterns so that we can decide which candidate should orshould not be generated by the next function. First of all, generate