You should learn SQL

Oct 22, 2016

I'm not the first to say this and I really doubt I'll be the last. Being able to use and understand data is becoming, if not already, a very powerful skill. Let me go through the two perspectives I've encountered below.

But I'm already learning Python/Pandas/R

Cool! So you're already interested and well on your way to being a data professional. Then there's two reasons you should learn SQL.

You'll become the data girl/guy faster

If you're trying to break into an analytical role your employer will be much more supportive, and likely even pay you, if you can show results. I would argue that SQL is the much quicker route to get a data role and start gaining experience for the following reasons.

Declarative vs Imperative Languages

Without going deeply into the nuances between declarative and imperative languages what's easier to write and understand?

import pandas as pd
df = pd.read_csv("http: iris url") # This is dependent on your data source
df.head()

SELECT TOP 5 * FROM IRIS

With Python you'll have know how to import pandas, call the correct method to read a csv file, know what argument to pass to the read_csv method, know about variables, and then know you can call a method on that assignment. A lot of the symbols are not "English" intuitive either.

SELECT TOP 5 * FROM IRIS

Comparatively, SQL is much more approachable and basically reads straight across. Now yes SQL has its own set of complexities and limitations but just starting out it's been my experience that there's much less you need to learn before you're able to do useful things. But....

It's probably already setup for you

I don't have strong evidence to back this claim I would argue nearly all businesses have a need to store data in tabular format and already do. This means you just need to start querying it, without having to worry about setting up a local Python or R environment, especially when most companies lock down admin privileges anyway and make this sometimes difficult to do.

Since it's setup there's a resident Database Admin

If the assumption above holds there already will be at least one person whose job it is to maintain that database who knows SQL. I would argue there's more companies that pay a DBA than companies that are hiring data scientists. In my experience nearly all companies have data these days, not nearly as many have Python/R stack data scientists on staff.

SQL has a standard

SQL is more standardized than other data languages. While there are differences the basic queries are pretty much the same across platforms. The switching cost between Python and R is much higher. This makes it easier to learn and use in a variety of settings, even if not every place you go uses the same database and SQL variant.

You're going to have to learn it anyway

Even if you end up using Pandas or R you'll have to find some data. While it's true a lot of data is not in database, a lot of data is in a database.

But I'm a manager

Even if you have direct reports that can write SQL, knowing a bit of SQL will help you become much more literate. Some of the managers who I've worked with have been able to use data to great effect to paint a story both to their direct reports and in discussions with their peers and superiors. It's harder to argue with the results from a query.

In Conclusion

If you're relatively new to data it's my opinion that learning SQL gives you the most "bang for your buck". It has a relatively low barrier to entry given that most companies already use it, and it's simpler to start out with and get useful results. And once your employer sees some results it'll be easier to sneak some Python/R and fancier stuff in as well!