Preppin' Data

Challenge by: Will Sutton We're starting June with a collaboration with #GamesNightViz and  #DataFamCon : a monthly challenge with topics covering a wide range of the geeky world including TV Series, films, books, and even some data about geek culture and conventions itself.  For this challenge, we'll be looking at a Dungeons and Dragons dataset from the podcast "Critical Role". If we wanted to know how long each character speaks for during each episode we'd need to do a little bit of preppin' to prepare the dataset for such an analysis.  There's an additional bonus challenge to build the following dashboard using the data we'll prep. Find the details at the bottom of this post! Find the viz designed by the super talented Christina Gorga here Inputs Episode Details  Dialogue  Requirements Input the data To create our gantt chart we'll need to work out how long each character is talking. To do this we can work out the difference from one timestamp ...

Challenge By: Jenny Martin Recently, my family and I have become quite invested in the TV quiz show Richard Osman's House of Games . The final round is always a round called Answer Smash. In this round you have a picture and question and you have to "smash" the name of the picture with the answer to the question, as per the below example. I thought it would be fun to work backwards and have a list of answer smashes from which we have to extract the Preppin' participant and the answer to the question! Inputs 1. Answer Smash list 2. Names 3. Questions 4. Categories Requirements Input the data The category dataset requires some cleaning so that Category and Answer are 2 separate fields ( hint ) Join the datasets together, making sure to keep an eye on row counts ( hint ) Filter the data so that each answer smash is matched with the corresponding name and answer ( hint ) Remove unnecessary columns Output the data Output 5 fields Q No Name Question Answer Answer Smash 20 r...

For many of the Preppin' Data challenges, we generate randomised data in order to keep the datasets freely shareable. Sometimes this can end up being more challenging than the challenge itself! If you've ever had to anonymise data so that you can publish a visualisation that you built using sensitive data, or built a dataset to work off of whilst waiting for access to the live data, then I'm sure you can sympathise with how frustrating this can be! You don't want the numbers to be so random that they appear unrealistic, but you also don't want to manually decide upon each value. Randomising the data at the lowest level of detail is one option, but can result in values varying too drastically when aggregating the data. A growth rate of 56% from one month to the next isn't always very realistic! This week's challenge is all about the process of generating the data for last week's challenge .   It's unlikely you'll get the exact same numbers s...

You can view our full solution workflow below and download it here ! The full workflow solution. This week we were working on recreating another table calculation – this time to moving average. This specifically example recreates the following Tableau Desktop settings: Tableau Desktop moving average settings. For each date in our data this provides us with a 1 week moving average which uses the current days sales & the past 6 days of sales and returns NULL for any date that doesn’t have at least 6 previous days. There are 4 stages to our workflow: Duplicate the data and subtract 6 days from each date. Join the duplicated days together so each date now has 7 rows; one for each day being used in the average. Aggregating to calculate the average. Adding NULLS where appropriate and rounding the results. Duplicating data and subtracting 6 days from each date The duplication and date calc section. In order to calculate our moving average we need to join th...