Answers

Here are the answers to the preceding questions:

1. No. We need to do a visual inspection of the data to guess the possible number of groups for the data. Then, we run the K-means elbow function to get the optimal number of groups of the data.
2. We choose the number where the curve starts to flatten.
3. The following are the parameters to execute the K-means function:
   • Number of groups to process
   • The range of the input data
   • The range to put the results returned by the K-means function
4. Use a pivot table and chart analysis with the minimum, maximum, and centroid values for each group. If the maximum and minimum values have a large distance from the average (centroid), that means that the group has scattered values.
5. Scattered values with a large distance from the centroid are possible outliers that need further research. One approach could be to do another K-means process for this group to create subgroups to improve the data classification.