Looking back on dashboard week, there’s definitely some things I’d go back and do differently, but I’m glad this is the case, because it means I’m making mistakes and learning. I’ve tried to whittle it down to the top three things I’ve learnt from dashboard week, that’ll hopefully be useful for me to reference back to, future DS cohorts and anyone else diving into the world of data.
1. Always try to read the supplementary documentation
This is something that I’ve not only realized during dashboard week, but during DS projects and my own individual work. When I joined the data school, I only looked at documentation when I needed to find a data dictionary or an explanation for a certain column. However, over the weeks and during dashboard week I’ve learnt that there’s a lot more useful information usually in this. The documentation can and often contains information on:
This is especially useful when dealing with survey data or data over several years. Did the data collection method change over time? Did the survey change? Is there missing data and if so, which method of data imputation was used?
Answers to these questions may not only help your analysis, but are also very important to make sure you’re not misrepresenting the data or not accounting for changes in data collection.
During dashboard week, Andy mentioned not to use all the data. This may seem obvious when you have a large dataset with many columns across different topics, but how do you decide on what to keep and what to drop. I found this challenging during dashboard week, where on a couple of days, I wasted time prepping all the data, to later only use a fraction of it. One method to avoid this is to pick your focus before prep, whether this be a specific question or subtopic within the data. Even then, I struggled to make a decision on this, feeling that I may regret it later.
Sometimes the documentation can help with this decision. For example, with my final dashboard on Freedom of Information Act Requests (see previous blog), I decided on efficiency of departments over time. Dropping data before 1998 made data prep and analysis quicker. I could make this decision more confidently after reading documentation that questioned the accuracy of data before this year. So, not only can reading documentation help to make decisions easier, not doing it could lead to inaccurate analysis later on.
Lastly, documentation may include links or direction to further reading or data to support your analysis. This may be a research paper or supplementary data that could help add context to your analysis. Both of these could lead to you understanding your data better, or may even spark some more ideas on what questions to ask during analysis.
2. Where possible, make decisions before data prep
This is much more specific to dashboard week than the previous lesson, but still useful to remember going forward. On a few occasions during dashboard week, I’ve wasted valuable time on unnecessary data prep on data I never ended up using. I touched on this in point 1, but even without documentation and under the time constraints of dashboard week, sometimes you just have to make a decision on what data to focus on. Applying this to my work in the future, I’m sure there’ll be times when a client presents me with work that I may have to choose from or prioritise, especially if the time frame is shorter than how long it’d take to complete all of the work.
3. Organise your files Joe!
Not a ground-breaking piece of advice, but probably a relatable one for many. During week one of the data school, I planned to be very organised with my file management, probably creating too many folders for all the different parts of training a file my fall under. However, this didn’t last too long. During DS training, you’re constantly creating new files for specific parts of training, and over the weeks I think I became slightly lazy, not naming files I wrongly assumed I wouldn’t need to come back to.
However, now I near the end of my training, and I naturally forget certain specific tableau tricks I learnt once two months ago, it’d be pretty useful to not have to spend 10 minutes trying to find that workbook that’s been named ‘Book1’. Generally, I’ve been better with organising project work, but as a general piece of advice to future cohorts, try to have some sort of file management and stick to it, because the seconds you think you’ll save by auto saving your work and not naming it, will just lead to you spending far more time frustratingly trying to find that file later in training.
Thanks for reading and I’m off to declutter and organise my laptop 👍