How to properly use the Tile Tool?

[jaipersr]

Hi I am having an issue with the Tile Tool. I have looked at the examples that come with Alteryx but, I am still having a problem. I am using the Equal Sum option to group records into tiles with equal sums. I have made an example case. Lets say I have a column with the following records in the following order:

 x

10

20 

30

40

50

60

70

80

Using 4 tiles my expectation would for the output using the x column as my sum field would be as follows. This makes sense to me since the sum of each tile is 90:

 x   Tile_Num  Tile_SequenceNum  

10         1                     1

80         1                     2

20         2                     1

70         2                     2

30         3                     1

60         3                     2       

40         4                     1

50         4                     2

However, in the output is actually as follows The sums for tiles 1, 2, 3, and 4 are 100, 50, 130, and 80, respectively. Any reason for this?:

 x   Tile_Num  Tile_SequenceNum  

10         1                     1

20         1                     2

30         1                     3

40         1                     4

50         2                     1

60         3                     1       

70         3                     2

80         4                     1

[PhilipMannering]

.

[DavidP]

Hi @jaipersr 

I'm a bit of a fan of the Tile tool myself, which is why your post caught my eye.

You're right, it does look like the equal sum option doesn't work at all.

The bit that's misleading in the explanation text in the example is underlined below:

In this output, the stores are tiled randomly and formed so that the sum of each tile's average sale columns should be equal to each other (this calculation comes as close as possible without breaking any rows).

It suggests that records would be paired in the most optimal way to produce X number of tiles with an equal sum, but as you duly found, this does not happen - records are tiled in sequential order.

The main problem with your dataset is that it's way too small. The result you got was the Tile tool's best effort at producing tiles of equal sum (in sequential order), and, it's fair to say that the results are a little disappointing.

To illustrate that I'm not making it all up, let's look at an example of 1000 records with values 1-1000 and do the same. With the help of the Summarize tool here are the results of the 4 tiles.

As you can see, the sums are pretty close and also note that the first 500 records are in Tile 1, the next 207 in Tile 2, 159 in Tile 3 and 134 in Tile 4.

[jaipersr]

Hi Thank you! That was my thought as well. I just wanted someone to confirm. Now I know to pay attention to the size of my dataset and that the Tile Tool groups records sequentially.

[ak2018]

Not entirely related to the question, I want to use the tile tool to create bins, I have values that I need to put into bins and therefore I use the Manual option in the Tile Tool where I specify the lower limit for a bin. 

The Tile Tool does its job just fine identifying the value and its bin. However, I want to assign labels without having to manually use the SWITCH or IF statements. Does anyone have any thoughts on how to achieve this?  TIA. 

[Alovasmith]

It sounds like you're diving into the complexities of the Tile tool, and you're absolutely right that the explanation in the example is misleading. The expectation of having the tiles' average sales sum equalized doesn’t fully align with how the tool actually works, particularly with smaller datasets. As you've pointed out, records are tiled sequentially, which doesn’t always lead to an optimal distribution for the equal sum calculation. When working with larger datasets, like the 1000 records you mentioned, the tiles can come much closer to equal sums, though the distribution still might not be perfect.

In this case, the Tile tool's performance is contingent on dataset size—with smaller datasets, like yours, the results can be less reliable. The tool’s “best effort” approach might work better with larger, more diverse sets of data, which provide more room for balancing the tiles effectively.

Thinking about your analogy with tiling and the need for precision, this is not too dissimilar to working with interior design. Just like achieving an even distribution of tiles in a space or ensuring the flow of a room, balancing data values within tiles requires thoughtful arrangement. An interior design specialist would approach the challenge of arranging a room with an eye for balance, flow, and utility, just as the Tile tool tries to balance data points. It’s not always perfect, but with the right considerations (like adjusting data size or using different tools), a better result can be achieved—similar to how a designer might rearrange elements to achieve a harmonious and functional space.

[Alovasmith]

To properly use the Tile Tool, follow these key steps:

1. Prepare Your Data: Ensure your dataset is large enough for optimal results. Small datasets may not work as well.
2. Select the Tiling Method: Choose between Equal Count (splits into equal record groups) or Equal Sum (balances values like sales across tiles).
3. Input Data: Make sure your data is sorted appropriately, especially if you're using the Equal Sum method.
4. Configure Tile Parameters: Adjust settings like the number of tiles or the field for tiling.
5. Test with Sample Data: Run the tool on a smaller sample to get a feel for how it balances the tiles.
6. Review Results: Check the distribution of tiles and their sums. If they’re imbalanced, tweak your parameters.

Just like an interior design specialist arranges a room for balance and flow, the Tile Tool organizes your data into evenly distributed segments. Whether arranging furniture or data, balance and thoughtful design lead to a harmonious outcome.