This webinar includes:
0:02:08 - Building an advanced analysis cube, using no code, in just 5 minutes.
0:13:52 - Demonstrating MODLR's natural language engine, to create reports and visualisations in seconds.
0:20:40 - MODLR's slack integration to enable team members to notify each other when data has been uploaded.
0:28:20 - New data spreading functions: Copy, Add, Subtract, Multiply, Divide.
0:35:37 - MODLR's new API integrations.
0:36:43 - New File Manager for uploading and managing data files.
0:37:42 - Data spreading - highlights.
0:38:20 - Advanced set instructions - highlights.
0:40:20 - Cube formulas - highlights.
Questions and Answers: AEDT and CST
In terms of the formulae that you demonstrated at the beginning of the presentation, how easy would it be to annualise a formula? I.e. if you're part way through a year and you want to compare it to prior year performance, how do you annualise YTD revenue?
If you have the standard Time dimension structure (or even a custom one) you should be able to apply a formula specifically to the YTD hierarchy and in it use the following formula:
IF(
ISLEVELZERO("Time")
,
CONTINUE
,
CONTOLIDATEAVERAGE("Time")*12
)
Making sure to select rate or average calculation which will ensure the calculation applies to the parent elements.
The IF statement makes sure that the calculation does not override any month level formulas.
Can you use elements from other dimensions in formulas?
Yes. The formulas themselves reference cells using square brackets and you can implement additional dimensional components to that reference, just by putting a comma in and then referencing the other element.
Can you have alternate hierarchies?
Yes. You can have any number of hierarchies within a dimension and a time dimension. All time dimensions, by default will have multiple standard hierarchies, for instance, year to date, a quarter view, a month view, and a month list hierarchy.
Is it possible to define measures of different types: text, date, time, array?
At present, it's just numerical and text. But you can use validations to create dates. For instance, when you're entering a date, it gives you a pre-predefined list of dates, that way, users can't go outside of that format.
Visualizations - NLP looks great. Is that out-of-the-box functionality? Does MODLR select the best chart type?
MODLR determines the best chart type, based on the key phrases in your query.
If I say:
"What's the trend of" - MODLR will create a stacked column and you can also configure that to use a line chart.
"How does something look over the last 12 months" - that will likely indicate that MODLR will create a trend chart.
"What's the top or bottom" - MODLR will automatically get a top x or bottom x bar chart.
"What's this versus that, or what's the relationship between these" - MODLR will create a scatter plot.
Building dimensions - error handling for orphans, double-counting, etc?
Yes. In the mapping process, if we created a circular reference, it would error out after 250 rows. So if enough rows have a problem in pulling in that data, it will abort that import, giving you the log of issues and will let you go back and amend those issues. For double counting, it will be harder for it to automatically detect that. You would need to go back and check your data to rectify a double counting scenario. Orphans shouldn't happen with how our mapping system works because you're telling MODLR every time you're creating a child, you're creating a parent for it as well.
Do you support unbalanced hierarchies?
Yes, MODLR does. So when you do a mapping process, MODLR will automatically balance because you're setting it at a fixed depth. But if you want to create it manually or if you create it using the scripts side of a process, you can quite easily engineer it, so that it will be unbalanced and the system will be fine.