✳flâneur — a map of the web's best reading

2. Exploratory analysis - Splink

moj-analytical-services.github.io · 583 words · saved by 1 readers

The purpose of exploratory analysis is to understand your data and any idiosyncrasies which may be relevant to the task of data linking. Splink includes functionality to visualise and summarise your data, to identify characteristics most salient to data linking. In this notebook we perform some basic exploratory analysis, and interpret the results. For the purpose of this tutorial we will use a 1,000 row synthetic dataset that contains duplicates. The first five rows of this dataset are printed below. Note that the cluster column represents the 'ground truth' - a column which tells us with which rows refer to the same person. In most real linkage scenarios, we wouldn't have this column (this is what Splink is trying to estimate.) Most of Splink's core functionality can be accessed as methods on a linker object. For example, to make predictions, you would call linker.predict(). We therefore begin by instantiating the linker, passing in the data we wish to deduplicate. It's important to

Exploratory analysis ¶ Exploratory analysis helps you understand features of your data which are relevant linking or deduplicating your data. Splink includes a variety of charts to help with this, which are demonstrated in this notebook. Read in the data ¶ For the purpose of this tutorial we will use a 1,000 row synthetic dataset that contains duplicates. The first five rows of this dataset are printed below. Note that the cluster column represents the 'ground truth' - a column which tells us with which rows refer to the same person. In most real linkage scenarios, we wouldn't have this column

Explore this link on the map →

related reading