Khipu

This project is about Bayesian spatial modeling of the Inca khipu dataset.

Background: Khipus are textile artifacts that Inca civilization used to record things. Their meanings are still not decoded yet. Our assumption is that khipus excavated from adjacent locations would be more similar than khipus from distant locations.

Approach: We use Bayesian conditional autoregressive (CAR) model to explore this spatial correlation. This is maybe the first attempt of using Bayesian statistical tool to analyze an archaeology dataset.

Results: With this model, we are able to impute locations for khipus which have unknown locations, compare regional difference of khipus, and possibly find interpretation of what khipus recorded.

Back to Homepage