Model Architectures11 min read• Published July 26, 2026

Tree-Based Architectures & Ensembles: Decision Trees, Random Forests, XGBoost, LightGBM & CatBoost

Impurity splitting criteria (Gini, Entropy), bagging variance reduction, and sequential gradient boosting algorithms.

GS
Govindarajan Selvaraj
ML Engineer

⚡ Executive Summary

Master decision tree split criteria (Gini vs Entropy), Random Forest bagging, and gradient boosted powerhouses like XGBoost, LightGBM, and CatBoost.

Key Takeaways

  • Decision Trees — covered in depth with practical examples, formulas, and code.
  • Bagging: Random Forest — covered in depth with practical examples, formulas, and code.
  • Gradient Boosting Frameworks — covered in depth with practical examples, formulas, and code.

Tree-Based Models & Ensemble Frameworks


#1. Decision Trees

Splits feature space recursively based on split criteria:

  • Gini Impurity: G = 1 - ∑ pₖ²
  • Entropy: H = - ∑ pₖ log₂(pₖ)
  • Key hyperparameters: max_depth, min_samples_split, min_samples_leaf.


    #2. Bagging: Random Forest

    Ensemble of independent decision trees trained on bootstrap samples with random feature selection. Reduces model variance without increasing bias.


    #3. Gradient Boosting Frameworks

    Sequential trees fitting negative gradients of the loss function:

    ModelSplit StrategyKey Strength
    XGBoostPre-sorted exact / approx splitsRobust, accurate, handles missing data
    LightGBMHistogram leaf-wise (best-first)10x faster training on massive datasets
    CatBoostSymmetric balanced treesSuperior out-of-the-box categorical feature handling
    python
    import xgboost as xgb
    model = xgb.XGBClassifier(n_estimators=200, learning_rate=0.05, max_depth=6).fit(X_train, y_train)
    Tags:#Decision Trees#Random Forest#XGBoost#LightGBM#CatBoost#Ensembles
    ABOUT THE AUTHOR
    GS

    Govindarajan Selvaraj

    ML Engineer

    Govindarajan Selvaraj is part of the Junglans Solutions engineering team, specializing in model architectures. Junglans builds a 20-product ecosystem of local-first enterprise software — AI developer tools, encrypted communication, and data infrastructure with zero cloud telemetry.

    Meet the full Junglans engineering team ↗
    RELATED RESOURCES & REFERENCES

    This article is part of the Junglans Research knowledge base, produced alongside the engineering teams that build our production AI tools. Explore related product documentation and research:

    Related Research & Articles