This blog goes through the solution for 2021: Week 25 - The Worst Pokémon Preppin' Data challenge to compare both Alteryx and Tableau Prep.
About the Challenge and the Data
The challenge required several joins to find the least favourite Pokémon in the anime series. Multiple tables were used in this challenge including Generation 1, Evolution groups, all Pokémon evolutions, Unattainable Pokémon in Sword & Shield, Anime appearances, and Mega evolution, Alolan, Galarian or Gigantamax form data. A total of 9 tables to work with =).
Luckily, a list of steps was provided by Jenny Martin who published this challenge. Click here to see the challenge description.
This section provides a summary comparison between Alteryx and Tableau Prep for this challenge in particular. More details for particular features and tools comparison between both software can be found further below.
- Based on my experience for this challenge, I believe that Alteryx is more suitable for larger or more complex workflows as it is easier to document and hence identify where you could have made a mistake. Nevertheless, Tableau Prep could be more suitable for quick and shorter workflows as it shows a more cleaner flow as long as the flow is kept short.
- In Tableau Prep it is easier to rename fields by just double clicking on it.
- I really like Tableau Prep's profile pane, data grid and list views are great to visualise the output and fields of each step. It is the equivalent to Alteryx's browse tool.
- In Tableau Prep it is easier to decide which join type to use, but I have to say that I personally find it easier to deselect fields in joins and visualise the output in Alteryx.
- There is no sorting in Tableau Prep and the rank is weird, although there is an option to rank records by selecting the field > create calculated field > select rank.
- I was unsure how to check for duplicates in Tableau Prep. In Alteryx you can use a Unique tool.
- The aggregation in Alteryx is easier to understand with the summarize tool than the aggregation in Tableau Prep.
Here is a table comparing certain features in both Alteryx and Tableau Prep based on this challenge:
|Renaming, excluding and keeping fields||Requires a select tool to exlude/include fields or dynamic rename to rename them||Double click on the field name to rename and right click on the field card to include/exclude|
|Filtering||Requires an easy-to-use basic or custom filter tool||You can create a filter or directly exclude/keep only a certain field or group of values by right-clicking|
|Cell split||Require text-to-columns tool but you have the option to split to rows in columns||Select custom or automatic split|
|Joins||In a normal join, you can only decide to have a left outer, right outer or inner join. You will need to union if you want to have a left or right join||Select which type of join you want from a Venn diagram|
|Unions||Union by name or position||I did not know if you could union multiple steps at once in the workflow|
|Cell split||Requires text-to-columns tool but you have the option to split to rows in columns||Select automatic or custom split|
|Sorting and Ranking||Sort tool available. Rank requires use of multi-row formula tool or related||Sort does not exist. You also need to use a rank LOD to rank values yet there is a calculated field option to select rank|
|Aggregation||Use a summary tool to aggregate the data. The output was a dimension grouped by Pokemon with the count of anime appearance next to it||Aggregation output gives you the grouped by Pokemon and count of appearance separate making it a bit harder to understand or visualise|
|Duplicates||Easier to check with a Unique tool||I did not know how to check for duplicates|
|Documentation||Easier to document workflow in Alteryx using containers or comment tool||I did not know how to document the workflow. You can rename the steps but if you have a large workflow it's hard to spot a certain part of it|
It was a fun and interesting challenge but it became really messy in Tableau prep due to the number of joins, unions and fields. Here is the final outputs from both methods:
Here is the final output posted in the challenge:
I have to say that the appearances in the final output don't seem to match the 'Anime Appearances' dataset which show way less appearances for some Pokémon such as Bellsprout. I also achieved 2 extra Pokémon in Alteryx - Weedle and Pidgey - which are also from Generation 1. I will update the post once I find out which solution is correct or more accurate but here is my final thought...
Alteryx, I choose you!