Tips and best practices
Use the following tips when working with the Assistant:
- Use natural language
- Use natural language for the best results when you use the Assistant. Like a real human, the Assistant attempts to understand your questions and queries based on the context of the whole sentence, including grammar, punctuation, and spelling.
- For example, if you asked your colleague for "can you show me my product profit?", they might
not understand if you're asking for two separate columns, called
profit, or if there's a column called
product profit. If you asked "can you show me my product and profit", then the question is clearer.
- Synonymous terms
- The Assistant also understands most synonymous terms and phrases. For example, show Profit is synonymous with list Profit, what is Profit?, tell me about Profit, and Profit.
- Synonymous terms also apply to many filters and aggregations. You can ask for the top 3 products, 3 best products, bottom 4 regions, 4 worst regions, how many products, show product count, and so on.
- You can also choose to spell out words or use symbols when applicable. For example, you can say show me sales greater than one thousand or show me sales > 1K.
- Load the correct data source
- When you are using the Assistant, the Assistant will identify the best data source for your question or query. After a data source is loaded, the Assistant sets the data source as the context for further questions. If a context is not set, Cognos Analytics uses the active data source in your dashboard or exploration panel as the loaded source.
- If you want to ask the Assistant questions or queries based on the context of a different data source, you can type load source <source name> to load the data source. You can also reset the conversation to load the active data source in your dashboard or exploration.
- Applying aggregations can help to add focus and create more compelling visualizations. You can use any aggregations that are supported in Cognos Analytics.
- Common numeric aggregations include:
count(used with measures)
- For categorical aggregations, you can use:
- Here are a few aggregation examples:
- show average <column-name>
- Displays the average for all values that are found in <column-name>.
- how many
- If <column-name> is a category, the number of distinct items is returned. If <column-name> is a measure, the count is returned.
- show maximum
- Displays the highest value found in <column-name>.
- show minimum <column-name>
- Displays the lowest value found in <column-name>.
- show total <column-name>
- Displays the total for all values that are found in <column-name>.
- show average <column-name>
- You can ask the Assistant to filter by using words and associated synonyms. You can add filters
for categories, including product columns, geographical strings (such as
Country or State), and temporal strings (such as
Month or Year). Here are some examples of how you can
- show the top 3 products by sales
- show products by sales where inventory > 500
- show products by sales for California
- show top <num> <column1> by <column2>
- Filtered visualizations include a filter icon in the chart. Hover over the icon to see the applied filtering. For more information, see Assistant filtering.
- Combining filters and aggregations
- Aggregations and filters can be combined to produce more granular results. Here are some
examples, based on sample data:
- show Revenue in 2017 and 2018
- show Income by Month for New York City, San Francisco, and Denver
- what are the top 5 States by average Inventory, excluding California
- show the average income, where income is less than 5K
- Retrieve your previous question
- You can use
up arrowto retrieve your last question.