ITDS Final

赶紧毕业吧

2024-12-10

Missing Values

Explain how would you pre-process the data if you would like to use linear classification/regression methods and the data would contain only categorical/nominal attributes. What could we do with missing values in this case?

解释如果您希望使用线性分类/回归方法，并且数据仅包含分类/名义属性，您将如何预处理数据。在这种情况下，我们可以如何处理缺失值？

Evaluation

Explain how internal and external evaluation of clusters work.

解释聚类的内部评估和外部评估是如何工作的。

Regularized LR

Write down the objective function for regularized linear regression. Explain, under which values (high or small) of the regularization hyper-parameter, the resulting model will overfit or underfit.

写出正则化线性回归的目标函数。解释在正则化超参数的高值或低值下，模型会过拟合还是欠拟合。

Error

What is the relation between the prediction error on the test set and the model complexity?

测试集上的预测误差与模型复杂度之间有什么关系？

Binary

What is a possible way to classify color images of animals to three different classes using binary classification methods? How would you represent the data? How would you do cross-validation in this case (i.e. how would you select the folds)?

使用二元分类方法将动物的彩色图像分类为三个不同类别的一种可能方法是什么？您将如何表示数据？在这种情况下，您将如何进行交叉验证（即，您将如何选择折叠）？

2024-12-27

Metric

Verify if $\mathbf{d(x,y)} = \max(|x - y|, 1)$ a distance measure for two binary strings $x$ and $y$ of equal length satisfies the properties of a metric.

验证 $\mathbf{d(x,y)} = \max(|x - y|, 1)$ 是否满足作为两个等长二进制字符串 $x$ 和 $y$ 的距离度量的性质。

Linkage

Provide a scenario or a dataset where complete linkage clustering would be less effective and justify your reasoning.

提供一个完整链接聚类效果较差的场景或数据集，并说明您的理由。

Equation

Given the following equation: $\sum_{i=1}^{n}(y_i - \hat{y}i)^2 + \lambda \sum{j=1}^{p}\beta_j^2$, what do the components of this equation represent? Discuss the impact of using very small and very large values of $\lambda$.

给定以下方程：$ \sum_{i=1}^{n}(y_i - \hat{y}i)^2 + \lambda \sum{j=1}^{p}\beta_j^2 $，该方程的各部分代表什么？讨论使用非常小和非常大的 $\lambda$ 值的影响。

Multi-Class

How can logistic regression be modified to perform multi-class classification?

如何修改逻辑回归以执行多类分类？

k-Fold

Explain the concept of k-fold cross-validation and describe the steps involved in performing it.

解释 k 折交叉验证的概念，并描述执行它所涉及的步骤。