Sharat Sachin Maximizing my potential

AWS ML Speciality (Part 3.1)

This post gives a quick review on the framing of business problems as machine learning problems.

Frame business problems as machine learning problems

Determine when to use/when not to use ML

Use ML when:

  • There is tons of data that need to be analyzed for patterns that can be used to make predictions
  • You cannot code the rules. No simple rule-based solution is possible
  • You cannot scale (say, to manually recognize a few million emails)

Don’t use ML when:

  • You want to solve a simple problem for which a rule-based solution would suffice
  • You don’t have labeled data and in-house expertise

Best practics when deciding on ML problem:

  • define success criteria
  • establish a performance metric
  • define inputs, outputs and metrics
  • decide if ML suitable?
  • data sourcing and annotation objectives
  • select a simple model

Know the difference between supervised and unsupervised learning

(From here)

Two types of classification: binary and multiclass

Selecting from among classification, regression, forecasting, clustering, recommendation, etc

Machine learning

  • Supervised learning
    • Classification
    • Regression
  • Unsupervised learning
    • Clustering
      • Topic modeling
        • LDA (Latent Dirichlet Allocation)\(^\star\)
        • NTM (Neural Topic Model)\(^\star\)
    • Embeddings
      • Object2Vec\(^\star\)
    • Anomaly Detection
      • Random Cut Forest\(^\star\)
      • IP Insights\(^\star\)
    • Dimensionality Reduction
      • Principle Component Analysis (PCA)\(^\star\)

(From here)

Sagemaker algorithms for each of them:

comments powered by Disqus