Data mining(I)

Learning Notes of Dr.Bo Yuan.THU 《Data：Theory and Algorithm》Part I

• Definition:Data Mining is the process of automatically extracting interesting and useful hidden patterns from usually massive,incomplete and noisy data.
  Not a fully automatically process.
  From data to intelligence.
  Data、information、knowledge、decision support
  

• Classification
  
  Algorithms:
  Decision Tree、KNN、Neural Networks、SVM
  Overfitting
  Cross Validation Training data 、Test data
  
  Confusion Matrix 、 TP(True Positive) 、FP（False Positive） 、FN（False Negative） 、TN（True Negative） 、TPR（True Positive Rate）、 TNR(True Negative Rate)、 Accuracy
  TP+FP+FN+TN = number of samples
  
  ROC:Receiver Operating Characteristic
  AUC:Area Under ROC Curve #AUC near 1 is good
  
  Cost sensitive learning
  Lift analysis

• Clustering
  Difference:Clustering is Unsupervised Learning,Classification is Supervised Learning
  
  Association Rule

• Regression
  
  Underfitting
  Overfitting

• Data Preprocessing
  
  Garbage Input garbage Output
  Cloud Computing
  Parallel Computing