Migration steps 1. Before migration: planning Planning is the best way to ensure the migration’s viability and that it will lead to the expected results in terms of data quality. Once the data quality objectives are clear, it is time to analyze the dataset and the source database. This includes answering questions such as: ● What is the volume of data to migrate? Is it a full migration or it is a partial migration? What are the criteria to select the data to be migrated? Does the data come from one or various sources? Are all the potential data sources identified? ● What is the situation of data in terms of quality (accuracy, uniqueness, timeliness, completeness, consistency and validity)? This analysis will allow us to identify quality gaps, but it is still required to assess if the gaps are solvable during the migration, and at what cost. Going back to the victim’s database example, the analysis could reveal that the phone field is empty for one third of victims, but the migration won’t be able to solve this problem. However, if duplicate victims are identified, it will be possible to reduce redundancy by eliminating one record or merging both. Data profiling The previous questions are part of data profiling, a set of techniques aiming to provide insights about the quality of the dataset. Data profiling allows us to quickly have a general overview of the dataset which often includes statistics, summaries of data types and patterns, blank values, etc. These are some of the most common statistics calculated for each column of the dataset (each column represents an attribute or property describing the entity): o Number of unique values and distinct values o Maximum and minimum values o Sum of values o Number of blank values o Mean, median, mode and range 7