If for me for a while, you may know that I started my work as Queer of Q Before converting to the world Data Analytics. I did not go to school about it, I didn't have a counselor, and I didn't get to a formal training program. All I know today – from SQL to model to data-teaching. And I believe, there was a trip, a mistake, learning, and re-reading.
The problem that changed my work
A few years ago, I began to think about changing organizations. As many people in strategic roles from soon, I faced a serious question:
What role do I really do? What roles should I apply?
On paper, I was Data analyst. But in fact, my role is suffered from several jobs: Writing SQL pipes, dash in KPIS, to describe KPIS, digging product. I was not sure that I had to apply for an analytical role, BIGS., Or something completely different.
Making things worse, at that time, the topics of Job were clear definitions, and work descriptions were flooded with buzzwords. You will receive a call called “Data Appup” Such written needs are similar to:
- Create ML pipes
- Write the complex text of the ell
- Keep data lakes
- Create Dashboard
- Provence Executive Insfights
- And oh, in a way, be great in the management of participants
It was difficult and confusing. And I know I'm not alone in this.
Top to today: Fortunately, things appear. It is still going through the roles, but the organizations have begun to explain them clearly. In this article, I want to break down Real difference between data rolesthrough the lens of the real example of the world.
The Real Hit Condition: Meet Quike
Let's think of Fant-Commerce Commercourn called Quikepresent to all many Indian cities. Their proposal of value? Move foods for purchase and importance in 10 minutes.
Customers put orders for the app or website. After the scenes, there are micro-warehouses (also called “black shops”) to all cities, and the delivery ships of the partners.
Now, let's go about the company's data requirements – from the order, Deshiboards management workers used at their morning meetings on Monday.
Step 1: Capture and retain green data
Currently customer places an order, Data for exchange produced:
- Vacation
- The order ID
- Things for the
- Price
- Discount Codes
- Customer Location
- Payment method
- A given assigned partner
Let's think to continue using Amazon Kinesis Streaming this data in real time in S3 Data Lake. That radio is the highest capacity, sensitivity for the time, and is important in accordance with business.
But here's a catch: The green data is dirty. You cannot use it directly to make decisions.
So what happens next?
Step 2: Building Data Pipes
Install Data engineers.
They are responsible for:
- To enter the original time data
- A valid SCHEMA agree
- Managing Failure and Recovery
- Pipeline to move data from S3 into data storage (means, Snowflake or Redshift)
That's where Interpretation (Uninstall, modify, upload) or Heart The pipes begin to play. Data engineers clean, format, and data component to be made from the shaped.
For example, the order table may be separated from:
- Orders → one line of each order
- Order_itites → one line per item by order
- Wages → One line to try to pay
At this stage, raw logs are answered by organized tables on colleagues.
Step 3: The model of size & olap
Since leadership begins to ask the formal questions such as:
- “Which city brings great revenues?”
- “Which store is not working well?”
- “What is our time for an average delivery of zone?”
… It is clear that the emergive transactions data will not be able to measure.
That's where Modensing a Figure Comes in.
Instead of low, green tables, details are organized on tables and regulations.
🔸 Facts Tables
- The big data tables, with a problem containing foreign keys and metric steps (Yes, most of the time. There are senseless and unreasonable tables).
- Examples:
fact_orders,fact_payments,fact_deliveries - Contain metrics such as income, order counting, delivery time
🔹 Tables in size
- Small tables, descriptive ones help understand the information on a particular table
- Examples:
dim_store,dim_product,dim_customer,dim_delivery_agent - Help sorting, group, and join facts with deep understanding
This building makes power Metal-Fast, analytical inquiry in the rest of the most size. For example, you can now run questions such as:
“Show me a regular time delivery time and the day for the past 7 days.”
This step is made from most of the data engineer of the organizations but form a few dim and true tables when working as an Amazon intelligence engineer.
Step 4: Explaining Kpis and metrics
That's where Analytics engineers (or engineers) shine.
They live in between the technology of the technical data and business users. Their works usually include:
- Description of KPIS (eg churn rating, repeat to buy%, time to fulfill)
- Writing logic of shiny metrics (eg Cohort maintenance, active users)
- Creation Models ASU or layers of thigh In tools such as DBT or abely
- To ensure consistent descriptions in the company
For example, Amazon, our team did not ask raw data regularly. Instead, we created True tables are combined in combined Things that are daily grain, weekly and monthly. That way, the Deserts are immediately uploaded, and the metrics live in all groups.
Analytics engineers serve as translators between engineering and descriptive business what are you We also measure How We measure.
Step 5: Analysis, Reporting & Talking Story
Now comes the role of Data analyst.
Equipped with clean, modified data, focused on answering real business questions such as:
- “Why is keeping down in Bangalore last month?”
- “What cups are driving very young users?”
- “Which high-editing products are edited in the first 30 days?”
They build Dashboard on tools like a table, power bi, or view. They run Ad-Hoc SQL questions. They enter the test results of A / B, the styles of the user's behavior, and campaign performance.
But above all, they Tell news With the numbers of making simple data be understood and applies too much to participants.
Who?
Tl; DR: where are you equal?
Here's what I think about it:
- Love strong pipes and love to solve solving problems? → UA Data engineer
- Love describes business metrics and planning complex datasets? → You Analytical engineer
- Love to kill understanding and to discuss matters with data? → UA Data analyst
Of course, the actual world roles usually mix these. Especially in smaller companies, you can wear many hats. And it's ok.
The key is not a title – but Where you enter the highest value including What gives you energy.
The last thoughts
It took me a long time to understand what I do actually – not just that my work title means. And if you ever felt a mess, you are not alone.
Today, I can clearly say that I work in the connection of Data Modeling, Logic of businessbesides Stories-The coloring between analysis and engineering. And I've learned that the power to connect the dots are more important than suitable in the full box.
If you go the same way – or wear lots of hats in your role – I would like to hear your story.
Pull Comments 👇 or share this with this person from getting again.
