I am trying to find outliers in a set of data. When I did the exploratory analysis, the outliers are clearly spotted in the histograms. Is clustering better than histograms? Do they provide more insight?
First, in a data set with more than one variable, you cannot really find outliers with histograms, since histograms are univariate and outliers can be multivariate (for example, in a survey of the American population, a 9 year old would not be an outlier and a widow would not be an outlier but a 9 year old widow would be an outlier).
Second, clustering can help find outliers, but its main function is to find clusters, often in data sets with several (or more) dimensions. Clusters are groups of observations that "go together" in some sense or other (with different clustering methods defining "go together" in different ways).