Additional Posts in Data & Analytics Consultants
1/15 Thread (BC):
My teaching philosophy in 5 words...
Conversation Starter
Additional Posts (overall)
Mentor
New to Fishbowl?
Download the Fishbowl app to
unlock all discussions on Fishbowl.
unlock all discussions on Fishbowl.
Bowl Leader
SQL is still used everywhere and is a prerequisite for almost all analytics jobs. It's also not coding in the traditional sense, you don't learn SQL as a first language "instead" of python, C++, Java, etc.
A non-coding way to start is learning a data visualization / dashboarding tool. Couple that with SQL and more advanced Excel, and you can apply to some business intelligence roles.
If you want to do data or business analytics, you'll need to learn some python or R as well.
BI roles definitely is more in-line with what I want to do - reporting and stuff using SQL + Tableau. Not sure about career progression for that though... never seen an analytics manager know only SQL/tableau & not python.
Thank you for your advice!
Coach
Not sure why you were told that. The vast majority of data is still stored in structured databases and you should know how to use SQL to extract it. That being said it’s not a standalone tool, if your data is just sitting in some csv you need another way to work with it.
If you don’t know the ML/stats you can start with either data viz and reporting, or data cleaning. You’ll spend most of your time on these tasks anyway, and they are a good way to start learning Python/R
Ok sure, but that still means you need to understand OOPs to use it...
Much like others have said, SQL is in no way shape or form outdated. Whoever told you that has no idea what they are talking about. Learn it, know it, because you will use it almost certainly.
For options, you could learn python (or R) or alternatively you could learn one of the many ETL tools that are on the rise today. I’m a huge fan of Alteryx personally. Think of it as “code-less” analytics though If you know code, the tool becomes significantly more powerful
SQL to learn the syntaxes for querying is a solid base. NoSQL is the infrastructure, but the syntax is similar.
Subject Expert
Already few useful suggestions. You could also try Product Analyst roles in Tech firms. Primarily involves SQL.
Also think of where you want to go long term. Don’t base your career on a piece of technology.
Product Analyst role actually looks perfect for my skill set and LT goals - thanks for that insight!
And lol yeah a little immature of me to avoid all roles/projects involving python, but it’s just so difficult for me to learn
Few related job titles who do little to no hands on coding: Data strategist, AI evangelist, Data monetization expert, Analytics story-teller, Big data enablement analyst
Most of those are just buzzwords LOL. Many D&A Director types don’t touch data though so many of the noted tasks - data strategy, evangelizing the AI/ML product, building decks that tell the story etc.
For the job that you mentioned you want to apply for, you should learn python even if you think it’s hard.
Look into data visualization (Tableau, power Bi). I don’t know coding and my job is strictly data visualization
Your job sounds awesome -I’ll connect
Why not become an analyst and specialize with products like Adobe/Google Analytics? The level of coding is simple and quite repetitive and you specialize on analyzing the data and provide insights on their performance. As others mentioned, you could also become an expert in visualization platforms or advanced Excel methods of data manipulation.
SQL is outdated? Who told you that. Maybe you are in the wrong field. SQL still has a good time to live and its easy to pick for sure but advanced concepts is truly what's needed!
How about learning to code?
Coach
1. SQL and Python is not considered code. They call it scripting. Find out why... if you know how to make excel do a vlookup, you can script.
2. With computers (hardware) being more accessible to masses, software (coding or no coding) will follow suite. If you want to future proof yourself, and not just in analytics, you should learn it.
3. Nothing worth doing is easy. If it is easy, you are underemployed.
Bowl Leader
Scripting vs coding is about use, not language. I don't care for the difference in terminology, but since some want to make hiring decisions from it, let's do this....
If all you do with Python is write one-off notebooks with no generalization, you are scripting. If you write anything at all as a class or a function, that's code, that's programming. Turing completeness doesn't guarantee anything, but does make it easier to write more types of programs. Lack of coherent syntax and community support makes it much harder to write "real programs," but it's done plenty.
To illustrate, let's consider the real controversial one, R. It's a high level Turing complete C based language with multiple object oriented systems and a quirk of just about everything being a list. For the vast majority of users, R is a scripting language. Built by and for statisticians, much of what's written in the language never makes it as a function, rather being saved as a global environment snippet and pasted script to script. Because of this, you'll often see job postings ask for both R AND a "programming language." However, that doesn't mean there isn't serious programming often being done in R. Every CRAN package is a piece of software designed to be distributed to thousands of users and built with some generalization. Internal deployments, though quite rare due to lack of extensive support from the bulk of the R community, do happen in R. I'll even make this personal: I wrote an ensembling enabled production ML framework in R, which is most certainly "coding."
Python is undisputably a first class (many say best in class) software development language. Many data science notebook scripts you may see are in fact scripts, but Python has a rich developer community, and as a language it is packed with developer conveniences. YouTube, one of the largest scale services on the planet, is written almost entirely in Python. In fact, this is why Python is the language of choice for data science at many companies: DS prototypes can be deployed with ease. On job postings, Python is often the sole requirement or listed twice: once as a data language, and once as a programming one.
Now SQL, the runt of the bunch. Though technically Turing complete, using ANSI SQL for anything other than data operations is outright uncomfortable and only done esoterically. It is a highly specialized language that's unreasonable to use otherwise due to a lack of basic programming structures developers take for granted, and thus you'll see it as a standalone line on job postings. ANSI SQL is a scripting language.
...but there's more. It is important to note that SQL's limitations are NOT because it is a declarative language. The catch here comes with SQL extensions, such as PL/SQL and TSQL. These are used to write packages used in databases, with code built to be generalized and maintainable. Dedicated engineering jobs like PL/SQL Developer are very real. Though these extension languages have syntax so specialized that they are cumbersome to someone who doesn't live and breathe RDBMS, they're commonly used to develop infrastructure rather than throwaway scripts. They are very much "coding" languages, though like Python, often make it into one-off scripts.
SQL is essential. Whoever told you that was clearly inexperienced. SQL is highly unlikely to be replaced in the next two decades. Sure if you prefer to learn a Python framework such as NumPy, PANDAS, or PyArrow, you can get away from pure SQL, but learning basic SQL is a few days of effort and worth the time.
Try qlik.com very easy to learn
Qlik can use sql but it's not recommended in most cases. Sql is linear . Qlik script is associative and super flexible. Easier too!
Visualization is part of almost every ML gig our team sells and can use little to no code depending on the tool. If you’re at M or D+ level, understand analytics and are good at storytelling, you can easily get by without touching code. I know 50+ that fit this type.
In previous years, not so much. Now, it’s a large part of nearly every ML project I’m on.