Conditional Attribute Values

Conditional Attribute Values are a tool that can be used to replace many existing transformers of the same type.

Transformer-Based Attribute Mapping

The Basic FME Desktop training course includes a section on Conditional Filtering, and explains how to apply conditions to data in order to split or divide it into multiple streams.

In this example, the AttributeFilter is being used to divide data on the basis of a postal code attribute value:

Similarly, a Tester or TestFilter transformer can be used to divide data in various ways.

So filtering transformers divide data up into separate streams in the workspace. The question is: what happens when the scenario requires a new attribute to be created based upon the original attribute value?

Technically, the following works, but it is far from elegant:

There are just far too many AttributeCreator transformers. Imagine if there were 100 values to handle! The workspace would be enormous!

In that scenario, the solution would be to use a simple AttributeValueMapper transformer:

However… what if the original attribute required a more complex condition to carry out the mapping? In that scenario something even more advanced would be required: Conditional Attribute Values.

What are Conditional Attribute Values?

Conditional attribute values are when the author sets an attribute value, but conditional upon a number of tests that must be first applied to the data.

The option for conditional attributes is found in the drop-down dialog on most transformer parameters. In the AttributeCreator, it appears like so:

The Conditional Definition dialog looks like this:

Like the AttributeValueMapper, a series of conditions map to different values. However, in contrast to the AttributeValueMapper, this dialog allows much more complex conditions than a simple 1:1 mapping. That’s because full test capabilities are built into this dialog.

The test capabilities are accessed by double-clicking in the Test Condition field. That opens up a dialog almost identical to that used by the Tester and TestFilter. A number of complex conditions can be defined in order for this test to be passed.

The output value for a passing feature can be set here at the foot of the dialog, or back in the Conditional Definition dialog:

When the conditions are set then the original dialog – in this case an AttributeCreator – looks like this, with the number of conditions defining the number of possible values:

When to use Conditional Attribute Values?

Conditional attribute values are great for when you need to map (or set) an attribute in relation to the value of an existing attribute, when the conditions are more complex than can be handled in a simple AttributeValueMapper or RangeValueMapper transformer.

A colleague is working on a workspace to calculate the flood risk due to tsunami for all addresses in the city. The risk is adjudged to be a combination of closeness to the shoreline and elevation above sea level, and is calculated using this table:

Your colleague has created the workspace up until the point at which the calculations need to start, and has asked for your assistance in finishing the project.

1) Start Workbench

Open the workspace

C:\FMEData2015\Workspaces\DesktopAdvanced\Exercise5b-Begin.fmw.

To find out what data we are dealing with, add Inspector transformers throughout the workspace and then run it - or else use Run > Run with Full Inspection from the menubar.

You’ll probably want to inspect the source features types (or the Reprojector, since the CDED data is a different coordinate system). You’ll also want to inspect the AttributeRenamer output. Don’t forget you can set a Group-By in an Inspector's parameters, which may be of use in visualizing which addresses are in which zone.

You’ll see how the addresses are assigned a zone denoting their distance from the shoreline, and also possess an elevation.

2) Assess Methodology

Before we get onto a full set of instructions for the exercise, try to consider how you might go about this task.

You’ll need to consider:

• Can the data be mapped directly, or does it need filtering first?

The zone is fairly easy to handle, because it is a fixed value (100, 200, and 300). However, elevation is trickier because they are not fixed values; elevation could be any single value from 0 to 60. So mapping elevation values would need 60 x 5 = 300 combinations!

• Which transformers would you use?

If you want to filter the data then the Tester and TestFilter transformers seem to be the most obvious candidates, with maybe the AttributeRangeFilter. To map data, either the AttributeValueMapper or AttributeRangeMapper appear to be the best. But as a whole, we’re looking to set attribute values, so why not just use the AttributeCreator?

• Should you combine methods?

Perhaps a combination method would work best, where you filter the data partially and then map it? If so, which data do you filter by and using which transformers?

• Which will produce the most aesthetic workspace?

Best Practice should always play a part in any workspace. If there are several solutions, then which produces the most aesthetic (good-looking) workspace? Are fewer transformers always better? Consider also the need to maintain the workspace. For the purposes of this exercise we'll assume that all methods have an equally efficient performance.

Try placing a few likely transformers into the workspace and see which might be suitable. Which will do the job, without needing too many transformers? Would conditional attribute values work?

3) Select Methodology

On consideration, I see three likely ways to carry out this project. I call them:

• Simple Filtering
• Complex Filtering
• Conditional Values

Simple Filtering filters the data and then maps it to the required values; as such it is a two-step process. It will require more transformers but will be simpler to understand and set up.

Complex Filtering filters the data in a single step, so no further mapping is required. It’s a onestep process, but – because the data is being filtered – still needs more transformers than perhaps necessary. It is moderately complex.

Conditional Values will set the values directly depending on a set of inbuilt conditions. All the work can be done in a single transformer, so it is compact, but the setup and maintenance are considerably more complex.

On the following pages we’ll detail how to set up each method. Pick which one you want to try and follow the instructions. Alternatively, do each in turn – then you’ll be able to compare the different methods and decide which you think is the best.

The Simple Filtering Method

This is a two-step process involving an AttributeFilter and several AttributeRangeMappers.

4) Place AttributeFilter

Place an AttributeFilter connected to the AttributeRenamer.

Open the parameters dialog.

Select Zone as the attribute to filter by.

In the Attribute Values field enter the values 100, 200, and 300.

You could use the Import function, but for so few values it’s hardly worth it.

Click OK to close the dialog and you’ll see a new output port added for each value you specified.

5) Add AttributeRangeMapper

Add an AttributeRangeMapper transformer and connect it to the 100 output port of the AttributeFilter.

Open the parameters dialog. As you’ll see this is a lookup table that involves ranges. We should be able to map the elevation range to a final flood risk using the information in the original table.

So, select Elevation as the Source Attribute. Enter FloodRisk as the Output Attribute

In the Range Lookup Table, enter the From-To values as follows:

If an elevation falls exactly on one value (for example 25) it will be counted in the lower band (i.e. 10-25).

In the Output Value fields, enter the values from the table for where Zone=100. These should be 1, 2, 3. Enter 999 as the Default, so that any features whose elevation does not match, for whatever reason, is flagged appropriately. Click OK to close the dialog.

6) Duplicate AttributeRangeMapper

Now we need to do the same thing for each of the other AttributeFilter output ports. Rather than set them up manually – as above – the easiest method is to copy the AttributeRangeMapper transformer that we just set up.

So, click on the existing AttributeRangeMapper and press Ctrl+D to duplicate it. Repeat and connect each duplicate to a different AttributeFilter output port.

The workspace will now look like this:

There are what we can call the 100m zone AttributeRangeMapper, the 200m zone AttributeRangeMapper, and the 300m zone AttributeRangeMapper.

However, what still needs to be done is that the Output Values need to be changed in each of these in accordance with the original table of calculations.

So, open the parameters dialog for each AttributeRangeMapper transformer in turn and change the output value fields.

The values will be: l 100m Zone: 1, 2, 3 l 200m Zone: 2, 3, 4 l 300m Zone: 3, 4, 5

7) Add Inspector

Place a single Inspector transformer and connect each AttributeRangeMapper output to it.

Open the Inspector parameters dialog and under Group-By select the newly created attribute called FloodRisk.

8) Save and Run Workspace

Save and run the workspace. You should see each address colored to match its flood risk. You can also turn off each zone in turn to see which addresses are most/least at risk.

The Complex Filtering Method

This also is a filtering process, but the filtering is all done in a single step - for both zones and elevation - with a TestFilter.

9) Place TestFilter

Place a TestFilter connected to the AttributeRenamer.

What we want to get here is a separate output port for each flood risk value. So we’ll need to incorporate all of the tests into this one transformer.

Open the parameters dialog. See that there are fields for Test Condition and Output Port.

Double-click the first Test Condition field and a Tester-like dialog will open.

This can be the test for FloodRisk=1 (the highest). According to the table of calculations, this can occur only when Zone=100 and Elevation <= 10.

So, set up the Tester to test for Zone = 100 AND Elevation <= 10. The important part here is to set up the test as an AND (i.e. both clauses) must be true.

Enter 1 into the Output Port parameter at the foot of the dialog:

Now click OK to close this part of the dialog.

The main TestFilter dialog now looks like this:

OK, now double-click the next Test Condition to set up the condition for FloodRisk=2

According to the table, there are two conditions for FloodRisk=2. They are when:

Zone = 200 AND Elevation <= 10

Zone = 100 AND Elevation <= 25

So, enter four clauses; one each for Zone=100, Zone=200, Elevation<=10, Elevation<=25.

Now change the test type to Composite. In the Composite Expression field, enter:

(1 AND 3) OR (2 AND 4)

Of course the composite expression field will depend on the order you entered the clauses in. If you entered them in a different order then you will need to adjust this field.

Enter 2 into the Output Port parameter and click OK to close this dialog. The main TestFilter dialog now looks like this:

Now repeat this step for each of the other flood risk values. There will be three clauses for zone 3, two clauses for zone 4, and back to one clause for zone 5.

It may seem complicated, but it should be easy to get into a routine. Additionally, make use of the Duplicate buttons in these dialogs to speed up the process.

The final dialog will look like this.

The final test will be similar to the very first, with only two conditions, so it will be an AND rather than a Composite test.

It is very important to keep these in the correct order; otherwise a feature may pass the tests in the wrong order and be given a lesser risk than expected.

The TestFilter transformer will now look like this:

10) Add AttributeCreator

Add an AttributeCreator connected to each TestFilter output port. Use the AttributeCreator to create the correct FloodRisk attribute (and value) for each output port (i.e. Port 1: FloodRisk = 1).

In fact, it’s probably easier to place one AttributeCreator and duplicate it for each port, editing the FloodRisk value each time.

11) Add Inspector

Place a single Inspector transformer and connect each AttributeRangeMapper output to it. Open the Inspector parameters dialog and under Group-By select the newly created attribute called FloodRisk.

7) Save and Run Workspace

Save and run the workspace. You should see each address colored to match its flood risk. You can also turn off each zone in turn to see which addresses are most/least at risk.

The Conditional Values Method

This is a one-step process involving an AttributeCreator transformer.

12) Place AttributeCreator

Place an AttributeCreator transformer and connect it to the AttributeRenamer.

Open the parameters dialog.

Under AttributeName enter FloodRisk.

Under Value click the drop-down arrow and choose Conditional Value.

See that there are fields for Test Condition and Output Value.

Double-click the first Test Condition field and a Tester-like dialog will open.

This can be the test for FloodRisk=1 (the highest). According to the table of calculations, this can occur only when Zone=100 and Elevation <= 10.

So, set up the Tester to test for Zone = 100 AND Elevation <= 10. The important part here is to set up the test as an AND (i.e. both clauses) must be true.

Enter 1 into the Output Value parameter at the foot of the dialog:

Now click OK to close this part of the dialog.

The main Conditional Definition dialog now looks like this:

OK, now double-click the next Test Condition to set up the condition for FloodRisk=2

According to the table, there are two conditions for FloodRisk=2. They are when:

Zone = 200 AND Elevation <= 10

Zone = 100 AND Elevation <= 25

So, enter four clauses; one each for Zone=100, Zone=200, Elevation<=10, Elevation<=25.

Now change the test type to Composite. In the Composite Expression field, enter:

(1 AND 3) OR (2 AND 4)

Of course this will depend on the order you entered the clauses in.

Enter 2 into the Output Value parameter and click OK to close this dialog. The main Conditional Definition dialog now looks like this:

Now repeat this step for each of the other flood risk values. There will be three clauses for zone 3, two clauses for zone 4, and back to one clause for zone 5.

It may seem complicated, but it should be easy to get into a routine. Additionally, make use of the Duplicate buttons in these dialogs to speed up the process.

The final dialog will look like this.

It is very important to keep these in the correct order; otherwise a feature may pass the tests in the wrong order and be given a lesser risk than expected. The main AttributeCreator dialog now looks like this:

13) Add Inspector

Place a single Inspector transformer connected to the AttributeCreator. Open the Inspector parameters dialog and under Group-By select the newly created attribute called FloodRisk.

14) Save and Run Workspace

Save and run the workspace. You should see each address colored to match its flood risk. You can also turn off each zone in turn to see which addresses are most/least at risk.

You’ll also see unmapped features appear, which doesn’t happen with the other methods.