Many Power BI datasets must be created without the benefit of a data warehouse or even a relational database source system. These datasets, which often transform and merge less structured and governed data sources such as text and Excel files generally require more complex M queries to prepare the data for analysis. The combination of greater M query complexity and periodic structural changes and data quality issues in these sources can lead to refresh failures and challenges in supporting the dataset. Additionally, as M queries are sometimes initially created exclusively via the Query Editor interface, the actual M code generated may contain unexpected logic that can lead to incorrect results and unnecessary dependencies on source data.
This recipe includes three practical examples of increasing the reliability of data import...