Intro to Data Mining (wk 11 Discussion reply 2)

Q 1 The first feature that indicates that the data is correct is accuracy. The integrity of the data is judged by how it is reflected in the real world. Completeness is another quality. The term “complete data” refers to comprehensive information. This entails gathering all of the pertinent information. Then there’s dependability. This indicates that the information supplied does not contradict itself. Finally, the data’s relevancy is quite important. The individual going into the field must understand why the data is needed. This will help them decide what kind of data to acquire and what data not to collect. Q 2 A prototype-based cluster is created in numerical data by calculating the average of all points inside the group. According to Martino et al. (2020), a prototype is considered medoid and rounded in the case of categorical characteristics. In irregular and interwoven collection, a graph-based cluster is employed. Two items are brought together with the requirement that a certain distance separates them. When there is a noisy, irregular cluster in a low-density zone, density-based is utilized. Q 3 A scalable cluster algorithm is a method for generating and collecting huge amounts of data (Alshehri et al., 2018). It can work with datasets that have both continuous and categorical features. This method may be scaled up to improve cluster number determination in noisy data. It’s also useful for determining cluster membership. Q 4 When selecting an algorithm, there are several crucial things to consider. The amount of training data must be substantial. This entails gathering enough data to produce a trustworthy result. A huge amount of training data resulted in more observations than features, requiring a high variance method. Another element is linearity. The majority of algorithms use the assumption that classes can be partitioned using a straight line. However, linear data is not required; logistic regression can be used instead. Requirements: 50 words