research

Research Folder

Contains research based components of the project including:

• dashboard_research.md provides initial research on dashboards with InfluxDB or Grafana to understand the level of effort associated with the task
• EDA folder contains the initial exploratory data analysis completed for the project
  • completed on a very limited dataset accessed through manual data downloads from the SkySpark UI
  • UDL_EDA.ipynb provides the notebook for the analysis
• model-methods folder contains the research associated with selecting the model to move forward with
  • LSTM model selected and comments on the selection are provided in the file model_selection.md
  • performance_selection.md provides comments on how we will evaluate the model/study
  • comparison_of_packages.xlsx provides comment on several packages that are available for anomaly detection in python (however, it was decided to build our own LSTM in Keras ultimately)
  • contains the model-demo folder which includes initial testing on various packages and the high-level tests on an LSTM model
• papers folder provides several references used in the project
• SR-testing testing spectral residual transformations on the manually labelled datasets
• streaming-methods folder includes various tests on streaming methods that could be used with InfluxDB
  • completed when trying to understand how the detection streaming framework with InfluxDB and what tools/packages would be required to implement it
  • contains README.md in the folder for description of findings
  • provides various .py and .ipynb files used for testing as described in the README