Report Details

Using “Knowledge of the Crowd” to Inform Subjective and Objective Projectile Point Clustering

(2021) Coal Creek Research, Inc.

Thirty-four volunteers sorted images of 259 projectile points into groups based on their own subjective preferences. Twenty objective parameters consisting of both metrics and proportions were also used to describe the projectile points. By analyzing the subjective groups using the objective parameters, nine of the objective parameters were identified as most related to the characteristics of the points used by each volunteer to make grouping decisions.

These nine subjective parameters as well as the twenty objective parameters were then used in three unsupervised clustering algorithms, and the reduction in the standard deviation of the parameters within groups were compared. These were also compared with the reduction in standard deviation of the parameters in the groups created by the volunteers. The clusters using the nine subjective parameters resulted in about twice the reduction in standard deviation compared to those resulting from the subjective grouping alone. This indicates the points in the objective clustering groups were more similar than those grouped subjectively by volunteers. Thus, objective clustering could produce both more repeatable results and groups that contained more morphologically similar points.

Document:

  Using “Knowledge of the Crowd” to Inform Subjective and Objective Projectile Point Clustering — (2021) Coal Creek Research, Inc.