How to Use the Model Builder

Overview

This video teaches you how to use the AI Model Builder to create your dimensions and data model directly from an uploaded spreadsheet. It shows how the wizard interprets your data, proposes dimensions like account, time, and scenario, and lets you confirm or correct its choices along the way. This is the fastest path to a working model without manually editing definitions.

Facts to Remember

• Account, time, scenario are mandatory dimensions in a data model.
• You can deselect value columns you don't want loaded before proceeding.
• If a scenario column isn't present in your source data, the wizard lets you name a default scenario (e.g. "Actuals") to map all rows to.

Transcript

All right, we're going to quickly create a model in Una.

To do that, we are going to talk a little bit about models and dimensions, and we're going to give you a quick little stop tour of doing this.

So obviously, we integrate to all different types of source data.

We go and grab that data, we pull it in, and we automatically build out.

Different types of templates and reports we call canvases.

But to do that, you need a model to start from.

So what is a model?

Well, the model is essentially how users interact.

It's usually combined of multiple dimensions.

Those dimensions are categorized sometimes with specific things like what's your account dimension?

What's your time dimension?

What type of scenarios do you have?

And what's great about it is you have an elastic set, meaning you can go as many dimensions as you want.

As few dimensions as you want, and you can actually have it automatically.

Connect to other different models, depending on which and how you decide to style and structure some of these.

So the core model is usually called finance or something like that.

And it will have a bunch of these dimensions.

As I mentioned, some of them will have roles and different types.

So different types of roles.

And other than that, it's really just to think about it as how would you slice and dice your data?

And so as you're integrating that data, you know, we normally want to also make sure that we're integrating the The data to automatically update these models as well, update these dimensions and update these components.

All right, so let's dive right into Una and start seeing how we can actually use the model building process to grab data.

All right, so inside of Model Builder, you're going to go and you're going to access some data.

You're going to go ahead and pull in a spreadsheet.

Typically, sometimes it's different data.

Sometimes it's usually using the data sources, data staging, data mapping.

So getting through those connections, but we'll start with an actual spreadsheet.

So to show you what this spreadsheet is, just here in Google Sheets, I'll show you what this representation might look like.

So I've got obviously some columns of metadata, or in this case, we'll talk about them in terms of dimensions.

So I've potentially got a product dimension, a pod, an account, a currency, et cetera, et cetera.

And I've got some values, what looks to be for the calendar year of 2024.

So we're going to use that data quickly here in Una.

To go ahead and build a model and deploy all of our default canvases.

So we're going to go and choose that file.

That pop-up doesn't exist in the recording, but it's gone and I've looked at my file system and then I've grabbed that data.

What the system is doing quickly is it will actually interpret all that it can in this and then it will start providing me a wizard to go through and finish refining how I want this model to react and behave.

So as you might have seen, it came back with a preview here and in this preview it's showing me just a few of those rows that you already saw in the spreadsheet and it's automatically started to guide me through what the next steps are, what dimensions we have and what it thinks it needs.

So the first thing it wants to know is it wants to verify that it got some information correct.

It got the information correct here.

On all of my different value columns, if I have more value columns, if I don't want to incorporate and load some of this data, maybe it's forecast data that I want to do just in Una, I can deselect these columns.

So anytime you have this ability to see that orange, it will allow you to go ahead and make some changes.

So we're just going to hit accept on those changes that it found.

You'll notice that I have already also Got the application, you know, when it does this AI build, it goes and looks at the columns and it tries to understand what's there.

So it's automatically showing me, here's my account name, here's my description.

It parsed that out.

You know, if I want to add things together, I can add things together.

I can use other AI capabilities here to actually go and drive what the caption should be.

So this is one of these scenarios where if you're using an ERP or a source system where you want to actually take out or add different things, you can do that right within the wizard.

So again, automatically.

Identifies things and then it gives you ability to go above this.

Now we can choose to upload different dimensional metadata.

So think of this as like parent-child of how things get represented.

Your income statement balance sheet and cash flow usually has a chart of accounts attached to it.

I'm just going to skip that for now and just see what the AI can do.

So the AI is going to now parse all of that account information that it sees and it's going to show me what this data is and you'll notice I automatically Selected to detect hierarchies, so it's going to go through and also see, hey, do I understand the pattern of this and what can it do?

So as mentioned, this system has gone and used the Una supervisor, called out to the chart of accounts interpreter, and it's looked at how does your account dimension naturally look.

So it's automatically grouped my revenue accounts that it sees, cost of revenue, and all the way down across my dimension.

So I have this ability to go down and see all that it's done.

I can add more members if I know that it's missing some data, but I'm just going to go ahead and click next.

It's going to then ask me, well, what time horizon do I want to work on?

So maybe I'm doing out another four or five-year plan.

Where is that data going to find itself?

It's going to find it in that first row.

So again, it's thinking very specifically about what it's been given, and it's trying to automatically guide you to what you should change.

Okay, so now it's created that, showing me the roll-up of all the different months, so I can say, yep, I'm happy with this. and it's going to go ahead and then say okay well what other dimensions do we likely need or have so we need a scenario dimension obviously we're going to put actuals in here i don't have a column for this so i'm just going to leave everything blank and the system is going to automatically ask me hey do you want to have your actual in here i could easily also put in my you know my budget or my plan So I can just create a name for that.

I can call it whatever I want here.

Obviously, if it sees that, it's now going to map to that, but the default member is going to be there.

So now, what of the dimensions do I want?

You know what, just for speed and expedience, I'm just going to deselect the pod and the currency.

This will allow me to move a little bit faster in our demonstration here.

And again, it finds the right columns for my various descriptions and names.

And it will allow me to go ahead and pull all these things in.

So I don't really care to have a hierarchy at this juncture for a couple of these dimensions.

So I'm just going to go ahead and make sure that I put that in.

So my department dimension, again, I can always choose if I want to, to automatically try to define a hierarchy.

It will do that all across time.

You can add more descriptive pieces to it so you can actually give it the way that it rolls up.

So I'm just going to keep going through here.

And now it's going to ask me for my product dimension here.

So again, it's going through the list.

It's saying, hey, I think I know what things are.

This one has the name and the column, just like everything else, right?

It's just going to go and do the work it can.

And then it's going to ask you for confirmation.

So we very much deploy a human in the loop view of this, of how we think about the world. so i'm just very quickly going in and adding all these things and now what you see is it's ready to actually load all of that data so it's showing me all the the tuples of information and based on that i can go in and build me my new model and what will happen now is the system will go ahead and you know start you know finishing the model development and actually loading that data so off we go

Related Articles

• Model Definitions — the settings area the Model Builder writes into
• Models — the underlying model definition, field by field
• Architecture — what the Model Builder is assembling underneath