Data Profiling

The first module of the pipeline. Data Profiling gives you a structured view of every variable in both the unharmonized and standard datasets before any harmonization begins.

Overview

For each variable in each dataset, aha automatically generates a description, identifies the data type, calculates the percentage of missing values, and gives the encoding details when applicable. Use this to understand the content and completeness of your data before proceeding.

Switching between datasets

The module displays one dataset at a time. Use the tabs at the top of the table to switch between the unharmonized source dataset and the standard dataset. Each has its own set of variables and profiling details.

The variable table

Each row in the table represents one variable. The table includes the following columns:

Column	Description
Variable Name	The name of the variable as it appears in the codebook.
Description	An extracted description from the data dictionaries of what the variable represents.
Data Type	The detected data type — e.g. `categorical_nominal`, `binary`, `text`, `numeric`.
% Missing	The proportion of rows in which this variable has no value.
Details	An eye icon that opens a panel showing the coded values for the variable, including any special missing-value codes.

Inspecting variable values

Click the eye icon in the Details column of any row to open a side panel. This panel shows the key-value pairs for that variable — the numeric codes and their corresponding labels — as well as any codes used to represent missing data.

Searching and sorting

Use the search bar at the top of the table to filter variables by name. The sort icon next to the search bar lets you sort by missing rate.

Where to go next

Once you’ve reviewed both datasets, aha moves to Strategy & Planning, where it proposes how each variable should be harmonized.