Compute Within-Cluster Sum of Squares
Arguments
- object
of class inheriting from
"kmeans".- newdata
matrix()new data to predict on.
Details
WSS is defined as $$ \sum_{i=1}^{n} \left\|x_{i} - \mu_{j(i)}\right\|^2 $$, where \(x_{i}\) is a data point and \(\mu_{j(i)}\) is the centroid of the cluster to which \(x_{i}\) is assigned. When new data is provided, the function predicts the nearest cluster for each new observation and computes the WSS for these points based on their predicted clusters.
Examples
km <- kmeans(mtcars, 5)
compute_wss(km)
#> [1] 4665.041 10247.471 8808.032 11474.702 7654.146
# or with new data
compute_wss(km, mtcars)
#> [1] 4665.041 10247.471 8808.032 11474.702 7654.146