Fault Diagnosis of an Industrial Chemical Process using Prinicipal Component Analysis (PCA) and Kernel PCA

  • Collected and pre-processed the Data
  • Performed Principal Component Analysis (PCA) to reduce the number of dimensions in the data
  • Chose the number of principal components to retain using the Percent Variance Test
  • Applied Kernel PCA and retained the principal components
  • Classified the faults in a supervised manner using Support Vector Machine (SVM)
  • Tuned the hyperparameters to select the best using GridSearchCV
  • Applied Recursive Feature Elimination to eliminate the least significant features in the dataset