3. ML模型类
3.1 Regression:
Linear Regression的基础假设是什么
what will happen when we have correlated variables, how to solve
explain regression coefficient
what is the relationship between minimizing squared error and maximizing the likelihood
How could you minimize the inter-correlation between variables with Linear Regression?
if the relationship between y and x is no linear, can linear regression solve that
why use interaction variables
3.2 Clustering and EM:
K-means clustering (explain the algorithm in detail; whether it will converge, 收敛到global or local optimums; how to stop)
EM算法是什么
GMM是什么,和Kmeans的关系
3.3 Decision Tree
How regression/classification DT split nodes?
How to prevent overfitting in DT?
How to do regularization in DT?
3.4 Ensemble Learning
difference between bagging and boosting
gbdt和random forest 区别,pros and cons
explain gbdt/random forest
will random forest help reduce bias or variance/why random forest can help reduce variance
3.5 Generative Model
和Discrimitive模型比起来,Generative 更容易overfitting还是underfitting
Naïve Bayes的原理,基础假设是什么
LDA/QDA是什么,假设是什么
3.7 其他模型
Explain SVM, 如何引入非线性
Explain PCA
Explain kernel methods, why to use
what kernels do you know
怎么把SVM的output按照概率输出
Explain KNN !所有模型的pros and cons (最高频的一个问题)
4. 数据处理类
怎么处理imbalanced data
high-dim classification有什么问题,以及如何处理
missing data如何处理
how to do feature selection
how to capture feature interaction