Automated machine learning

History / Edit / PDF / EPUB / BIB
Created: December 7, 2019 / Updated: December 7, 2019 / Status: in progress / 2 min read (~307 words)

  • This assumes a supervised problem such as
  • Classification
  • Regression

  • Determine needs
  • Collect datasets
  • Clean datasets
    • Remove invalid rows
    • Impute missing values
    • Correct data
  • Merge datasets
  • Create dataset relevant to the problem
  • Generate features
  • Define the metric to optimize
  • Train and evaluate various models on the provided dataset
    • In certain domains, evaluate the model against a baseline
  • Analyze the produced predictions
    • Worst offenders analysis
  • Generate a list of improvements suggestions
  • Prioritize the improvement suggestions
  • Implement the improvement suggestions