Frosi Exam: Chapter 4

Wat er op dit rad staat (39 opties)

• What are we focusing on with unsupervised models?
• Introduce clustering and the advantages over classification and regression.
• Describe one of the three algorithms discussed in clustering.
• Name the output of clustering. Where to cut off the dendrogram and the purpose of using the silhouette approach.
• What is one of the most common approach to clustering?
• How would you use top bottom approach? Bottom up?
• What are some issues with k-mean?
• Briefly explain the components on which DBSCAN is built on.
• How would you select for k? What parameters affect k-means?
• Describe the bottom half of class clustering?
• Do you remember the task that is the most time-consuming? Why is it important?
• What are core points? Boundary points? Outliers?
• Describe bottom up clustering.
• Describe what clustering is.
• K-Means. How does the initial selection of the centroid influence the resulting clusters? Do you want them close or far apart?
• What are we focusing on with unsupervised models?
• What are clustering useful for?
• What does a good quality clustering look like?
• What is the goal of clustering?
• What are some applications of clustering?
• What are the 2 main types of clustering methods?
• What are the resulting clusters from hierarchical and partitioning clusters?
• What is bottom up approach? Pros and cons.
• What is top down approach? Pros and cons.
• What are the types of linkages found in hierarchical clustering?
• What is K-means algorithm? Pros and cons.
• What happens when you increase k in k-means?
• What pre-processing tasks must you do before using k-means?
• What post-processing tasks must you do before using k-means?
• What are some tips and tricks when using k-means in the initial assignment of points?
• What is density-based clustering? Name some algorithms?
• What are the parameters of density-based clustering?
• Define directly density-reachable, density-reachable and density-connected?
• What is DBSCAN algorithm? Why does it fail sometimes? Pros and cons
• How do we evaluate clustering?
• What are the metrics to evaluate clustering and to be classified into?
• How can be SSE do used to evaluate clustering?
• What are some of the internal indices?
• What is Silhouette Coefficient. Average silhouette? When and where would we use this?