Alfred 4 tutorial series#

It places the data from both data frames into one series.

Internally, the SortedNeighbourhoodIndex does the following to determine candidate links: However, knowing how this is processed internally is important for understanding when this method might return unexpected results. This shows how the sorted neighbourhood method can be used to identify candidate links even when you cannot guarantee exact matches between potential matches. # Create pandas MultiIndex containing candidate linksĬandidate_links = indexer.index(df_a, df_b)Īs can be seen in the network diagram of the result, we have candidate links between all names which would be adjacent in a sorted list. Indexer = rl.SortedNeighbourhoodIndex(on='names', window=3) Names_2 = ĭf_a = pd.DataFrame(pd.Series(names_1, name='names'))ĭf_b = pd.DataFrame(pd.Series(names_2, name='names')) The interleaved neighbourhoodĬonsider the following code, which creates a set of candidate links using the sorted neighbourhood method:

Alfred 4 tutorial series#

This tutorial contains a series of examples, which are intended to demonstrate both the power and the quirks of the sorted neighbourhood method. To get the most out of this tutorial, you should already be familiar with indexing candidate pairs with recordlinkage. This tutorial takes a closer look at the sorted neighbourhood indexing method, which has more complex behaviour than other indexing methods. This tutorial is a supplement to our tutorial on indexing candidate pairs with the recordlinkage Python package for data integration.