Please enable JavaScript.

Coggle requires JavaScript to display documents.

Model Development - Coggle Diagram

- - - - Data Cleaning
        
        Feature Selection and Engineering
        
        Model Selection
        
        Metrics
        
        Model Training
        
        Model Tunning
        
        1 more item...
        
        Technical metrics
        
        Business metrics
        
        Linear
        
        Non-Linear
        
        Mutual Information
        
        Correlation
        
        Create feature
        
        K-mean cluster
        
        1 more item...
        
        Missing Value
        
        Duplicated Record
        
        Outliers
        
        Scaling
        
        Normalization
        
        Flooring and Ceiling
        
        Parsing Dates
        
        Character Encoding
        
        1 more item...
        
        Get year, month, day, week
        
        Standardization
        
        0 and NaN: do they have meanings?
        
        Train test split before imputation
      - Continous
        
        Categorical
        
        Time series
        
        Unstructured text data
        
        Embedding, bag of words, word2vec, etc
        
        Check Stationality, Seasonality, ACF, PACF
        
        For cyclical features (e.g. Week, Month), use sin+cos feature engineeringg
        
        Add Lag, Rolling means as features
        
        Dummy, One-hot, Label encoding
        
        Make sure train and test has the same categorical encoding. Train-test split before encoding
        
        Mutual information, Correlation
        
        Train test split before feature engineering
  - - - Categorical
        
        Classification
      - Continuous
        
        Regression
        
        Not normally distributed? Consider log transformation