![]() ![]() Second, Q&A makes some educated guesses about how users are likely to talk about date and geography columns to help it understand certain types of questions. It assumes that results containing columns with geographical Data Categories might look good on a map. For example, it recognizes that columns with date or time Data Categories are a good choice for the horizontal axis of a line chart or the play axis of a bubble chart. Q&A uses this information in two important ways, for visualization selection and for language biases.įirst, Q&A uses the Data Category information to help make choices about what kind of visual display to use. For example, you can mark an integer column as a Zip code or a string column as a City, Country/Region. The Data Category provides knowledge about the content of a column beyond its data type. This setting is in the Properties section after you select a column.Ĭhoose a Data Category for each date and geography column Other columns that aren’t sensible to sum, such as Age, could also benefit from setting the default Summarization to Don't summarize or to Average. Be mindful of Year, Month, Day, and ID columns, as those columns are the most frequent problems. If you have specific columns where you don't want Power BI to exhibit this behavior, set the default Summarization property on the column to Don’t summarize. Power BI aggregates numeric columns by default, so questions like “total sales by year” can sometimes result in a grand total of sales alongside a grand total of years. Select the correct data type in your Power BI model.Ĭhange the year and identifier column settings In particular, date and number columns that are imported as strings aren't interpreted by Q&A as dates and numbers. Imported data can have incorrect data types. Tables named Store and Products work better. Table names like StoreInfo and Product List need work. ![]() It would be better to rename those tables to truly reflect what they contain. This result is probably not what they had in mind, because it’s a count of every job every employee has ever had. Someone else who asks “count the employees” is going to get a count of the rows from the “Employees” table. People familiar with the model might understand this structure. You have another table named Employees that contains employee numbers, job numbers, and start dates. While Q&A can do some basic word breaking and detecting plurals, Q&A assumes that your table and column names accurately reflect their content.Īnother example could be if you have a table named Headcount that contains first and last names and employee numbers. You would need to ask a question like “List the customer summaries in Chicago” rather than “List the customers in Chicago”. For example, say you have a table named CustomerSummary that contains a list of your customers. The choice of tables and columns is important for Q&A. In the second image, relationships are defined between the tables. In the first image, there are no relationships between the Customers, Sales, and Products tables. The following images show a model that needs work and a model that's ready for Q&A. For example, you can't ask for the “total sales for Seattle customers” if the relationship between the orders table and the customers table is missing. Relationships are the cornerstone of a good model. If your model is missing relationships between tables, Power BI reports and Q&A can't interpret how to join those tables. For more information, see Intro to Q&A tooling. In this case, read the following sections to help you optimize Q&A. Sometimes, questions still can't be addressed because the data is shaped incorrectly or data is missing. With Q&A tooling, you teach your core business terms to Q&A and fix questions your end users ask. In the following sections, we describe how to adjust your model so it works well with Q&A in Power BI. Those adjustments for Q&A are the same best-practice optimizations for any model in Power BI, regardless of whether you use Q&A. If the structure of your model doesn't meet one or more of these assumptions, you need to adjust your model. To enable Q&A to successfully interpret the large collection of questions it's capable of responding to, Q&A makes assumptions about the model. It's even more powerful when your data answers, which is what the Q&A feature in Power BI does. It's powerful to use common phrases and natural language to ask questions of your data.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |