If you’re a data analyst or considering becoming one you may think you don’t need to learn Data Science. As someone that was first a data scientist and then a data analyst, I never realized how useful it would be knowing how to build machine learning models when I became a data analyst. Today I’d like to discuss why a data analyst should consider learning data science even if they don’t plan to become a data scientist.
Increase your chance of getting hired
Junior analyst jobs are competitive and any additional skills you have can stand out against candidates applying for the same position. Looking at LinkedIn this data analyst job for Time shows over 3,700 applicants while this data analyst position for Blue Bottle has over 2,500 applicants within a week of getting posted.
Responsibilities for mid and senior level data analyst jobs may require you to know how to build models. Companies may post data scientist jobs containing a majority of data analyst responsibilities and add building models as another requirement. If you learn how to build basic Machine Learning models this will qualify you for these type of hybrid positions and justify a higher salary compared to data analyst positions that do not require modeling experience.
For example, this Twitter mid-level data scientist position doesn’t mention building models as part of the job responsibilities but asks for an applicant with "2+ years of work experience in data-driven analytics/predictive analytics/statistical modeling/machine learning/data mining" and bonus points if you have "experience building ML models". This Atlassian senior data scientist position doesn’t require modeling in the job responsibilities but under "it’s great, but not required, if you have" it mentions "experience using statistical and machine learning methods to build descriptive and predictive models".
You may lose out being selected for senior level jobs if you don’t know the data science basics.
Using data science for data analysis
A machine learning model can provide a shortcut to finding predictors of user behavior. For example, a mobile product manager wants to know what causes a user to continue using the app versus those that stop. As a data analyst you may analyze app events by user to determine which events show the highest retention rate. Now imagine how time consuming it would be if you had 100 events to analyze. If you use a machine learning model to predict a user’s app retention you can get a list of top events that cause a user to come back to the app. This may cut down the list of events to analyze from 100 down to 10 because the model showed the remaining 90 had no impact on user retention.
You don’t have a build a model to predict the future you can use the model output as a guide on where to look deeper.
I once had a take home assignment for a data analyst position that stated "An executive producer at a movie studio has given you this dataset, and wants you to give a brief overview of an interesting insight derived from it that will help the studio’s business." I used a machine learning model on the movie data to find multiple actionable insights instead of just one insight the assignment requested and was able to find the insights faster.
How I Used a Machine Learning Model to Generate Actionable Insights
Expand your career options
I mentioned earlier there are data scientist title jobs that include building machine learning models but the majority are data analyst responsibilities. You may prefer this type of hybrid role if you don’t want to build models full-time but like working on data analyst projects. Alternatively, you can start off as a data analyst but change your mind and decide to become a data scientist. If you already spent time as a data analyst learning data science you’re closer to building out your skill set to apply for a data scientist position.
There are already many skills you need to learn when you first start out as a data analyst and you may decide it’s not worth spending additional time to learn data science. I hope now that you know the benefits, you’ll reconsider learning data science even if you don’t plan to become a data scientist.