What is a key characteristic of the k-Nearest Neighbors (k-NN) algorithm when used in data mining?

Question

BlackTom AI · Accepted Answer

k-NN is an instance-based (lazy) learning method that stores training examples and postpones generalization until a query is made.

1. Relevant knowledge points and concepts
- Instance-based (lazy) vs. model-based (eager) learning: lazy learners store data and perform computation at query time; eager learners build a compact model during training.
- k-NN mechanics: store training set, compute distances (e.g., Euclidean), select k nearest neighbors, use majority vote for classification or average for regression.
- Non-parametric method: no fixed set of parameters learned by optimization; memory- and compute-intensive at query time.
- Common distance metrics and issues: Euclidean, Manhattan; sensitive to scale and curse of dimensionality.

2. Step-by-step reasoning process
- During "training" k-NN simply stores labeled examples (no model fitting).
- When a new query arrives, k-NN computes distances between the query and each stored example.
- It selects the k smallest distances (nearest neighbors) and aggregates their labels to produce a prediction.
- Because prediction requires these on-the-fly computations, the algorithm effectively delays prediction until the query — hence “lazy learning.”

3. Example calculation (all intermediate steps)
- Training points: A(1,1), label = Class1; B(2,2), label = Class1; C(5,5), label = Class2.
- Query Q(2,1). Compute Euclidean distances:
  d(Q,A) = sqrt((2-1)^2 + (1-1)^2) = sqrt(1 + 0) = 1
  d(Q,B) = sqrt((2-2)^2 + (1-2)^2) = sqrt(0 + 1) = 1
  d(Q,C) = sqrt((2-5)^2 + (1-5)^2) = sqrt(9 + 16) = sqrt(25) = 5
- For k = 2, nearest neighbors are A and B (both Class1) → prediction = Class1. All computation happened at query time.

4. Why the correct answer is correct
- The defining characteristic of k-NN is that it does not build a general model during training; it waits and uses stored instances to answer each query, matching the given statement exactly.

5. Why the other options are incorrect
- "Requires training a complex model like neural networks": false — k-NN stores instances, doesn’t train multi-layer models.
- "Always uses gradient descent to optimize parameters": false — there are no parameters to optimize via gradient descent in basic k-NN.
- "Builds a decision tree based on entropy or information gain": false — that describes decision tree algorithms (e.g., ID3/C4.5), not k-NN.

6. Conclusion
The correct answer is: It is a lazy learning algorithm that delays prediction until a query is made — this matches the provided answer.

What is a key characteristic of the k-Nearest Neighbors (k-NN) algorithm when used in data mining?单项选择题

类似问题

Which of the following is true of the value of k in the k-Nearest Neighbors algorithm?

The algorithm is a classification scheme that attempts to find records in a database that are similar to one we wish to classify.

Question11 (Blank A) is a non parametric and instance based Machine Learning algorithm.Select what best fit (Blank A) Linear Regression Decision Trees k-Nearest Neighbour Random Forest ResetMaximum marks: 0.5 Flag question undefined

给定一下列表：You are given the following table: 如果应用 kNN 算法， k=1，对数据项 8 的决策是什么？If apply kNN algorithm, what is the decision for data item 8?

最近邻分类器 k-NN 是基于什么来指定标签？ Nearest Neighbor Classifier k-NN assigns a label to a node based on

更多留学生实用工具

智能学习助手

风格化写作助手

论文查重助手

文献引用助手

课堂转译助手

课堂笔记助手

Quiz搜索助手

学校历年真题

智能刷题助手

智能匹配练习

希望你的学习变得更简单