Please enable JavaScript.

Coggle requires JavaScript to display documents.

BigQuery for Data Warehousing - Coggle Diagram

- - - - Artificial Intelligence
        
        Computer to do human like stuff
      - Machine Learning
        
        Use data to predict or classify
      - Statistics
        
        More on why the data looks the way it looks, ML models are built to produce outputs
      - Ethics
        
        Implicit Bias
        
        Disparate impact
        
        Responsibility
    - - Cost
      - Supervised vs Unsupervised Learning
        
        Supervised
        
        Labelled data
        
        Input
        
        Desired Output
        
        Regression
        
        80% Training, 20% Validation
        
        Another testing split with data is also typical
        
        Unbiased data the model does not see at all
        
        Unsupervised
        
        No labelled data
        
        Coz no specific desired outcome
        
        Search and extract relationships
        
        then produces clusters or detect anamalies
        
        BQML natively supports k-means clustering
        
        Useful in identifying patterns that are unseen to human observers
      - Model Types
      - Procedure
    - - K-Means Clustering
      - Data Classification
    - - ML with SQL
      - Future - Crossover between Data Science and Traditional BI