課程摘要
透過集合功能整理內容
你可以依據偏好儲存及分類內容。
以下簡要說明您在課程中學到的內容：
- 決策樹是一種模型，由一組條件組成，並以樹狀圖的形式進行階層式排序。條件可分為以下幾類：
- 訓練決策樹時，系統會在每個節點中搜尋最佳條件。splitter 例程會使用資訊增益或Gini等指標，判斷最佳條件。
- 決策樹是一種由多個決策樹組成的模式。決策樹的預測結果是決策樹預測結果的匯總。
- 隨機森林是決策樹的集合，其中每個決策樹都會使用特定的隨機雜訊進行訓練。
- Bagging 是一種技術，可針對隨機森林中的每個決策樹，分別訓練不同的範例子集。
- 隨機森林不需要驗證資料集。相反地，大多數隨機森林會使用稱為out-of-bag-evaluation的技術來評估模型品質。
- 梯度提升 (決策) 樹狀圖是一種決策樹狀圖，透過輸入決策樹的迭代調整進行訓練。名為「收縮」的值會控制梯度提升 (決策) 樹的學習速度，以及過度擬合程度。
參考資料
除非另有註明，否則本頁面中的內容是採用創用 CC 姓名標示 4.0 授權，程式碼範例則為阿帕契 2.0 授權。詳情請參閱《Google Developers 網站政策》。Java 是 Oracle 和/或其關聯企業的註冊商標。
上次更新時間：2025-02-25 (世界標準時間)。
[null,null,["上次更新時間：2025-02-25 (世界標準時間)。"],[[["\u003cp\u003eDecision trees utilize conditions organized hierarchically to make predictions, with training focused on finding the optimal condition for each node.\u003c/p\u003e\n"],["\u003cp\u003eDecision forests combine predictions from multiple decision trees, while random forests introduce randomness during training to improve performance.\u003c/p\u003e\n"],["\u003cp\u003eRandom forests employ out-of-bag evaluation for model assessment, eliminating the need for a separate validation dataset.\u003c/p\u003e\n"],["\u003cp\u003eGradient boosted decision trees are iteratively trained with adjustments influenced by shrinkage, balancing learning rate and overfitting potential.\u003c/p\u003e\n"]]],[],null,["# Course Summary\n\n\u003cbr /\u003e\n\nHere is a quick summary of what you learned in the course:\n\n- A [**decision tree**](/machine-learning/glossary#decision-tree) is a model composed of a collection of [**conditions**](/machine-learning/glossary#condition) organized hierarchically in the shape of a tree. Conditions fall into various categories:\n - An [**axis-aligned\n condition**](/machine-learning/glossary#axis-aligned-condition) involves only a single feature. An [**oblique\n condition**](/machine-learning/glossary#oblique-condition) involves multiple features.\n - A [**binary\n condition**](/machine-learning/glossary#binary-condition) has two possible outcomes. A [**non-binary\n condition**](/machine-learning/glossary#non-binary-condition) has more than two possible outcomes.\n- Training a decision tree involves searching for the best condition at each node. The [**splitter**](/machine-learning/glossary#splitter) routine uses metrics such as [**information\n gain**](/machine-learning/glossary#information-gain) or [**Gini**](/machine-learning/glossary#gini-impurity) to determine the best condition.\n- A [**decision forest**](/machine-learning/glossary#decision-forest) is a mode made of multiple decision trees. The prediction of a decision forest is the aggregation of the predictions of its decision trees.\n- A [**random forest**](/machine-learning/glossary#random-forest) is an ensemble of decision trees in which each decision tree is trained with a specific random noise.\n- [**Bagging**](/machine-learning/glossary#bagging) is a technique in which each decision tree in the random forest is trained on a different subset of examples.\n- Random forests do not require a validation dataset. Instead, most random forests use a technique called [**out-of-bag-evaluation**](/machine-learning/glossary#out-of-bag-evaluation) to evaluate the quality of the model.\n- A [**gradient boosted (decision)\n tree**](/machine-learning/glossary#gradient-boosted-decision-trees-gbt) is a type of decision forest trained through iterative adjustments from input decision trees. A value called [**shrinkage**](/machine-learning/glossary#shrinkage) controls the rate at which a gradient boosted (decision) tree learns and the degree to which it could overfit.\n\nReferences\n----------\n\n- [Greedy Function Approximation: A Gradient Boosting\n Machine](https://jerryfriedman.su.domains/ftp/trebst.pdf), J. Friedman.\n- [The Elements of Statistical\n Learning](https://web.stanford.edu/%7Ehastie/Papers/ESLII.pdf), Trevor Hastie. Chapter 10.\n- [Generalized Boosted Models: A guide to the gbm\n package](http://www.saedsayad.com/docs/gbm2.pdf), G. Ridgeway"]]
