IDA

IDA

1. PCA

   • Expectation
   • Variance

2. Mining Text Data

   • doc Vectors
   • Clustering/Classification
     • Why?
       • data sparse
       • Synonmy&Polysemy
     • How?
       • LSA
         • PCA
         • Concept Space
           • Semantic relationship
             • co-occurence of terms

3. Page Rank

   • Energy

4. Clustering

   • codebook vector
     • initialization
       • iterative method
   • How many
   • Where

5. Classification

   • K-NN
     • Majority Vote
     • Two extremes
       • K = N
         • too simple
       • K = 1
         • too complex
       • Best K
         • precision in classifying my data
         • stability across mutiple sample
   • K-fold