Please enable JavaScript.

Coggle requires JavaScript to display documents.

MACHINE LEARNING (Problems (Data Leakage (Tips to combat Data Leakage…

- - - - model that uses the response variable itself as a predictor, thus concluding for example that “it is sunny on sunny days.”
      - “customer service rep name” feature variable in a SaaS company churn prediction algorithm.
    - - 1. Perform data preparation within your cross validation folds.
        
        :warning: Do not normalize or standardize your entire dataset
        
        In general though, it is a good idea to re-prepare or re-calculate any required data preparation within your cross validation folds including tasks like feature selection, outlier removal, encoding, feature scaling and projection methods for dimensionality reduction, and more.
        
        Platforms like R and scikit-learn in Python help automate this good practice, with the caret package in R and Pipelines in scikit-learn.
      - Hold back a validation dataset for final sanity check of your developed models.
    - - Clean dataset
      - impute missing values
      - remove outliers, etc
    - - Temporal Cutoff.
      - Add Noise.
      - Remove Leaky Variables.
      - Use Pipelines.
      - Use a Holdout Dataset.