From the course: Learning Data Analytics: 1 Foundations

Discovering common beginner mistakes

From the course: Learning Data Analytics: 1 Foundations

Discovering common beginner mistakes

“

- [Instructor] People make mistakes. I wanted to share with you where I recognize most mistakes being made by new data analysts. When you're new to a role, you definitely want to impress, but this can be a dangerous misstep if you do not spend time learning about the data. Anyone can make a Pivot Table show up or create calculations, but did you pay attention to how that data is structured and what it meant to the overall objective? No matter how small the dataset or how large and complicated, please spend time looking over what the field headings are, what the data types may be and understanding of the values. People not looking for duplicated data in a set is a major issue. Again, if you spend a little bit of time up front, you'll be able to spot those. However, the bigger the dataset, the harder it is to detect duplicates. It's impossible to detect them if you're not looking for them. There isn't a single dataset that I work with that I do not sum the numbers, average them and count them. It gives me a lot of information up front that I need later to verify my records. Record counts of each dataset that you work with are invaluable to you as you begin to work with the data. If you're joining data with queries, you must know the record counts to validate your dataset. If I have 50 product records and 1,000 sales records and I'm producing a sales order report, there's never going to be more than 1,000 sales. And if all the sudden I see 50,000 sales records, I immediately know I have an issue. Time is valuable and it's a non-renewable resource. When you know that your time is limited with decision makers be sure you have your questions documented ahead of time. That way you know all the questions you need to cover. Also document the answers you've received, and follow up with confirmation that you have the correct understanding. I think one of the scariest things for me as a data analyst is knowing that anyone can make numbers show up and that they might appear to be correct. I would ensure before you go to the annual meeting and you've just told everyone that this is the best year of record, or the worst ever, that you've done some form of verification of your calculations. You want to make sure that what you're showing is correct. From the newest analyst to the most senior, you must verify your numbers and calculations. Keep in mind that logical errors do not produce an error message. They're just incorrect. So take that extra time to verify your work. It's hard though, because sometimes people have been encouraged not to ask questions. I believe that this happens based on the types of questions people have asked over the years. However, if you have critically considered your question and you know you need an answer from an authority, then you definitely need to ask. Don't guess the answer. People will often provide you what they think you need but not always everything you need. Be sure that you've asked for documentation. It could be that no one has ever asked for that documentation before. If it exists, you'll work smarter and not harder. And if it doesn't exist, then you've lost nothing. And if you create it, then you're providing even more value than just the data. I think sometimes people think that they just need to jump in, and they don't spend enough time analyzing existing reports. There is literally nothing wrong with looking at system reports. I did it for years, and in fact I encourage people to do it to save time. The people who built the system are typically responsible for what we call canned reports. There's a lot of value in understanding how the developers joined all the pieces together. Also reviewing the reports of others, if they're willing to share. Understand if they're not sharing, it's likely the fear of someone else peeking over their shoulder. People are thrust into report writing regardless of skills. So it may be they're hesitant to share because they're scared of what you'll think about their data. Just carefully ask for information that might be useful to you as a new person to this data. If you can avoid mistakes by leaning on what others have already learned over time, you will definitely come in more confident about your work overall.

Contents