It is an unfortunate truth that human intervention is still required to properly cleanse data within a business. It would be nice to rely on Machine Learning or some other Automated solution, and we have little doubt that the technology will get better and better at doing this. Indeed in the five years that Power BI has been in existence, the value of machine learning has increased exponentially; the challenge, in this case, is, however, that at some point, you need to able to completely turn it over to the system or you will need to manually validate. In practice, manually validating unique values in a single column is a relatively quick process. Power BI’s value is that we can automate the steps we need to automate knowing that the dataflow or Power BI will follow the steps or let us know if there is an exception. While a column like “Lab Name” manually entered has so much variation that any poor attempt to automate will require constant human interaction to be valid, acknowledging that, at least for now and set up a manual management plan is the most efficient way of working.
IIn the last two episodes, we’ve covered ways of extending analysis and using Time Intelligence. Still, we saw last week that we needed to do some more work to get to all the support measures written for a new measure, so when we created a “Business Rules” version of “Inspection Count”, we had to also create YTD, Last Year, AVG Year etc… using Calculation Groups we would have had those measures available to use whenever we wanted. There would be no requirement on you to write all those extra measures, only the basic version.