Visual Analytics Meeting



    • 12:00 PM 12:30 PM
      Introduction 30m
      Speaker: Maria Grigoryeva (National Research Centre Kurchatov Institute (RU))

      Need to discuss:

      1) After-conference results
      2) Intel DAAL KMeans instead of sklearn/KMeans in InVEx LoD clusterization (add virtual environment setup for daal4py library in twiki)
      3) Log Messages Clusterization Django App
      4) Implement string clusterization at LoD level in InVEx
      5) K-neighbors implementations benchmarks
      6) Features selection for clusterization algorithms

      TO DO:
      1) NEC Proceedings - 20 November! ( @matitov, @archangel153 )
      2) Fix saving of state at Clusterization left panel ( @archangel153 )
      3) Move InVEx at conda with installed daal4py ( @matitov )
      4) Add IntelDAAL KMeans in LoD and Clusterization Left Panel ( @matitov )
      5) Implement the first stage of the HDF5 data model ( @aleksee1 )
      6) Design HDF5 file structure with LoDs ( @maria_a_grigorieva )
      7) Clusters treemaps + statistics tables on Par-Coords page ( @archangel153 )
      8) Implement K-neighbors benchmarks, select the best one and include it in error-logs-clustering library ( @dgrin, @maria_a_grigorieva )
      9) Implement pip-installed library for error-logs-clustering ( @maria_a_grigorieva )
      10) Add nearest neighbors plot to help users to choose epsilon for DBSCAN ( @dgrin )
      11) Left Panel: clusterization - issues with a number of clusters and clusterization in general
      12) Add the ability to choose normalized/non-normalized data 
      13) Add PCA algorithm for dimensionality reduction at the LoD level

      1) display operation info on Three.js canvas panel - Add dat.gui panel with info about visualized clusterization
      2) [after database] develop GUI to navigate user operations - LoD, clusterization, grouping etc. : multilevel drop-down list (pure JavaScript)
      3) add Progress Bar to LoD
      4) Dynamic page titles for InVeX ( @tanyakorchuganova, @matitov )