Last updated on 2023-05-01 | Edit this page
- What is Clustering in OpenRefine and when would you use it?
- How does clustering work in OpenRefine?
- Explain what clustering is in OpenRefine
- Use clustering to identify and replace varying forms of the same data with a single consistent value
The Cluster function groups together similar, but inconsistent values in a given column and lets you merge these inconsistent values into a single value you choose.
This is very effective where you have data with minor variations in data values, e.g. names of people, organisations, places, classification terms.
To use the ‘Cluster’ function, click on the
menu option in the relevant column and choose
Cluster and edit...
The ‘Clusters’ are created automatically according to an algorithm. OpenRefine supports a number of different clustering algorithms - some experimentation may be required to see which clustering algorithm works best with any particular set of data, and you may find that using different algorithms highlights different clusters.
For more information on the methods used to create Clusters, see https://github.com/OpenRefine/OpenRefine/wiki/Clustering-In-Depth
For each cluster, you have the option of ‘merging’ the values together - that is, replace the various inconsistent values with a single consistent value. By default, OpenRefine uses the most common value in the cluster as the new value, but you can select another value by clicking the value itself, or you can type the desired value into the ‘New Cell Value’ box.
- Clustering is a way of finding variant forms of the same piece of data within a dataset (e.g. different spellings of a name)
- There are a number of different Clustering algorithms that work in different ways and will produce different results
- The best clustering algorithm to use will depend on the data
- Using clustering you can replace varying forms of the same data with a single consistent value