As a student in the Master of Science in Business Analytics program at UC Davis, we get to have many hands-on experiences with data analytics, such as projects and competitions. One of the experiences that I think I have learned the most from is my practicum experience. In a team of five students, we got to work with the company BlueMatrix to help the company dig into its data and bring valuable insights through analysis. We got to work with real-world problems and do analysis with true purposes.
Through this practicum experience, there are three things I have learned that I think will help me become a better data analyst in the future. In the analysis flow mentioned in the book Analytics Body of Knowledge edited by James J. Cochran, my three recommendations will mainly fall into the stages of Data and Analysis. We will begin with understanding and exploring the data, then move on to things to pay attention to in the analysis and insight stage.
Understand Your Client
The first thing that every analytical project should start with is understanding your client. There could be three perspectives: the business, the needs, and the capabilities. In our case, BlueMatrix is a company in the financial research industry, which was an industry that we did not have much domain knowledge in. It was important for us to understand what the critical metrics in the industry are, and what are some trends that we should look out for. Just like the article “6 Easy Steps to Help You Understand Your Client’s Business” mentioned, the first step is to “Get a grasp on your client’s brand and industry trends.” We conducted research on the internet and spoke with the Chief Technology Officer of BlueMatrix to gain an insider view of the industry.
Further, for the company itself, we had to speak with different departments to understand how the company functions. This will help with understanding the flow of information and focus the analytics on the most important parts of the business. Here are some questions that you could ask when trying to understand a business:
1. Who are the targeted customers?
2. Where does the revenue come from?
3. What is the customer journey when they use the product or service?
After having some knowledge in the industry and the operation of the business, we proceeded to understand the needs of our client. Our project was a pretty open-ended one, thus it was critical for us to narrow down the scope of the project to take action. We had conversations with people from the marketing team and the sales team, the CTO, and the product managers. One important thing for these discussions is to come with questions and suggestions, so the conversation can be more directional and end up with better results.
Lastly, analysts have to understand the tools and capabilities of the company. For example, we decided to make an interactive dashboard for BlueMatrix to monitor and track key performance metrics, and we wanted to use Tableau in the beginning because that was the tool we were most familiar with. However, when we brought up the idea at our weekly meetings, we realized that BlueMatrix didn’t have a license with Tableau at this point, and to integrate the dashboard into their current system, the best way was to produce it in Python. So, in the end, what you are most comfortable with is not important when comparing to what your client is most comfortable with.
Get your hands on the data before doing analysis
It is a common practice to conduct Exploratory Data Analysis (EDA) at the beginning of the project to get an overall sense of the data. Here is how IBM discusses its importance:
“The main purpose of EDA is to help look at data before making any assumptions. It can help identify obvious errors, as well as better understand patterns within the data, detect outliers or anomalous events, find interesting relations among the variables.”
We have found this to be extremely helpful to us as well. The dataset we have received consisted of many tables and A LOT of information, so it was difficult for us to connect the dots in the beginning. The first thing we decided to do is to build an Entity Relationship Diagram and highlight the fields that we think will be most useful for us. This helped us map the available information and formed a better understanding of what was available.
When we did our EDA, we also tried to go beyond the typical contents in an EDA report and tried to provide some basic insights on the data. For example, instead of simply reporting the mean and median of the research readerships, we decided to plot them on a time series plot and highlight some trends we observed there. Then we went further to explore the topics that BlueMatrix was particularly interested in.
When you get your hands on the data and play with it before start doing the actual analysis, you have a better sense of the larger picture of the data and can detect some potential risks such as data accuracy issues that can be mitigated ahead of time. For example, if the company you are working with started collecting data in the year 2000, but the dataset contained records from 1900, then you know there must have been an issue there.
Working under limited resources
One of the most inspiring things I have learned through this experience is how to deal with limited resources or work under time constraints. Our mentor told us about this trilateral relationship that we should consider in this situation: Scope, Accuracy, and Delivery on Time. When we have enough resources and an unlimited amount of time, we can try to take as long as we need to do the best we can in all three aspects. However, you don’t always have that ideal situation. In those cases, you can try to give up one of them.
For one of the projects we worked on, we did not want to give up the accuracy and we wanted to deliver on time, so we decided to narrow down the scope of our analysis to the most recent two years. It is easy for us to fall into the trap of trying to make everything perfect, but sometimes choices have to be made and it is ok to make them.
As the practicum and learning continue, I am sure that I will gain more practical insights on how to become a better data analyst. Experiences build on each other, and I am very thankful to my program for offering the opportunity.