DEV Community: vanshika1807

Understanding the Big Picture: AI to DL

vanshika1807 — Mon, 06 Jul 2026 15:00:59 +0000

Suppose there is a Universe & we call it AI.

The ultimate goal of many AI projects is to build intelligent AI-powered applications.

AI refers to systems that can do its task on its own with little to no human intervention.

Example -

Recommendation System
Self Driving Car
Amazon -> If you buy something, then other similar or useful things are automatically recommended

Machine Learning is a subset of AI. It provides us the stats tool to analyze, visualize data & then it will be helpful in forecasting or prediction of new data.
DeepLearning is a subset of ML.

How can we teach ML to minimize human efforts?

We use multi layered Neural Network for DL! We train our ML algo!

Where does Data Scientist fits here??

They are the part of everything out of AI/ML/DL, they combine domain knowledge & uses Probability/Statistics/Linear ALegbra/Linear Regression to collect data, analyze it, build predictive models, and solve real-world business problems.

The goal is to generate valuable insights and, in many cases, build intelligent AI-powered applications.

We solved & encounter only 2 type MAINLY

But lets talk about dataset first.

We usualy have dependent(input) & independent variables (target/output)

here, we are taking example of 2-D means 2 features

Taking example of Weight & Age

We will having a basic data set that we will using to train our model. So, We will make a ML algo that takes any unknown age and will gonna give a predicted the weight.

This is nothing but Hypothesis. We use Hypothesis to actually make that straight line.

Usually, We know about the output feature like in our case the output weight (dependent variable)

In supervised learning, we solves two types of problems: Classification & Approximation.

For more details, may be you can refer this article of mine on Classification & Approximation: Classification v/s Approximation

In Unsupevised Learning, the dataset that have no labeled output. Instead, the model tries to discover hidden patterns or groups within the data.

What we do is, we use the most common method called clustering & based on those cluster we build our models.

These clusters can then be used for customer segmentation, recommendation systems, anomaly detection, and many other real-world applications.

Data Science is not equal to just Models

vanshika1807 — Sun, 05 Jul 2026 04:22:06 +0000

vanshika1807

Jul 4

From Confusion to Clarity: Understanding Data Science Techniques

#datascience #ai #computerscience

4 min read

Data Science is not Just Models

vanshika1807 — Sun, 05 Jul 2026 04:21:03 +0000

vanshika1807

Jul 4

From Confusion to Clarity: Understanding Data Science Techniques

#datascience #ai #computerscience

4 min read

Data Science Just Models

vanshika1807 — Sun, 05 Jul 2026 04:21:01 +0000

vanshika1807

Jul 4

From Confusion to Clarity: Understanding Data Science Techniques

#datascience #ai #computerscience

4 min read

From Confusion to Clarity: Understanding Data Science Techniques

vanshika1807 — Sat, 04 Jul 2026 19:52:52 +0000

This article is simply an attempt to organize everything I've been learning in one place.

When I first started learning data science, I was overwhelmed by the number of techniques, algorithms, and buzzwords. Every new topic seemed to introduce another family of methods!!

The list goes like beloww

regression, classification, clustering, dimensionality reduction, recommendation systems, neural networks... the list just kept growing.

I'm writing this from the perspective of someone who has only recently stepped into this field. Yes, we're already living in the AI era, and sometimes it feels like I'm very behind. But I've realized that one of the best ways to learn is to document the journey.

So, I started asking myself a few simple questions:

What exactly are we trying to solve?
Why are there so many techniques in data science?
Why do we need dozens of algorithms for seemingly similar problems?
How do data scientists decide which technique to use?

The answer begins with understanding the problem, not the algorithm.

Many predictive machine learning tasks can be grouped into two broad categories:

Classification – predicting a category or class (for example, whether an email is spam or not).
Regression (Function Approximation) – predicting a continuous numerical value (for example, forecasting house prices).

However, data science is much broader than prediction alone. We also encounter problems such as clustering similar customers, detecting fraudulent transactions, reducing the dimensionality of data, forecasting future trends, recommending products, processing text, analyzing images, and much more.

Before choosing any technique, we first need to understand the data itself.

In the real world, data is rarely clean or perfectly structured. Businesses generate massive amounts of information every second!

From application logs and server metrics to customer transactions, clickstreams, sensor readings, and social media activity. In many cases, this isn't just "data"; it's Big Data.

Some datasets are structured and easy to work with, while others are noisy, incomplete, inconsistent, or completely unstructured.

That's why the first job of a data scientist isn't to build a machine learning model but it's to understand the data and the business problem. Once both are clear, we can identify the type of problem we're solving and select the most appropriate family of techniques.

In this article, I'll explore these different categories of problems and the techniques commonly used to solve them.

Classification Problems

Here, We deal with labled data, where the goal is to assign a label or class to a new data point based on learned patterns. Classfication also further have several types.

Binary Classification-

Has only 2 label. you have data that is already labeled. Our Model is Trained on it. Now, New data point will come and the new data point will a unlabled data. The class of that data will be pridicted based on the features it has and will be given a type.

Example - Fraud Detection

Two Classes

Every transaction has various features such as

New Data comes in X

Based on the featurs it will be given a class either Fraud or Legit.

Multi-Class Classification

Example - Real Engineering System often requires distiniguishing among several classes, such as failure modes.

Here, We will be having several classes like Fault1, Fault2, Fault3

Same idea but have more labels. Historical Data tells us: when these values looks like 'THIS',the pump was fine. When look like 'THAT', it was F2.

New, Data comes in, algo says 'this was 85% F2', we alert the maintainance team even before the pump/machine breaks.

Linear v/s Non-Linear

This about the shape and boundary that seperates the classes

Now, which technique should we pick??

Function Approximation

We have continous output/data. In this case, we have a function that maps input with output.

Predict tomorrow's server reponse and it will predicted based on load, time of day, active users etc,

this data is continous

Linear Case (Linear Regression)

Non-Linear Case

So, in Non-Linear case which shape should we pick to represent the data. That is where it become complex

Generally, a data can have many attributes(100 attributes) it means 100-D, we cannot imagine that , we cannot plot that, We can't see patterns.

So, What do you do? You make assupmtions!

Data is seperable --> Linear classification

output is continous --> certain stat model

There are natural clusters --> Use K-Means

The Loop-

When nothing seems to stick

vanshika1807 — Sun, 22 Mar 2026 12:09:09 +0000

There are versions of tech industry that nobody talk about

Not like cracked a Tier-1 Company in few months or "Created a start-up while in college"

Tried everything, mastered nothing

At one point, I thought the solution is simple I need to learn more

So I started jumping between thing:

Java
Python
Kubernetes
Data Analytics
Cyber Security
Cloud
DevOps

Yes, I know this is very absurd & stupid but it is what it is!

Every week felt like a new direction.

I would start something, get into it, and then suddenly feel:

“Is this even the right path?”
“What if something else is better?”
“Am I wasting time?”

So I’d switch again.

Not because I was lazy (I am though >.<), but because I didn’t want to go in the wrong direction.

The Hidden Problem -> No Clear Guidance

There is too much out there & to much to focus on, so many videos out there that says Do DSA , Focus on building Projects, Build your Tech-Stack, Learn Cloud, Go Into AI because it is the future, SRE is the future, then DevOps is the future.

Everyone sounds confident.
But no one tells you what you should actually do.

And when you’re early in your career like me, that noise gets overwhelming.

Cherry on top - A new layer enter - AI

just when the thing were already confusing AI entered the picture.
Everywhere I look, I see:
"AI will replace developers"
"IT industry will collapse soon"
"There are no jobs for entry level associates"
"You’re already late if you’re not using AI"
"One person can now do the work of ten"

And it adds a new kind of pressure.

Not just:

What should I learn?

But also:

Is what I’m learning even going to matter?

What am I learning about this?

Honestly, I do not have an answer.
I have started seeing thing differently

Did I go deep enough to understand how things actually work?

I cannot learn everything out there, there always be something left. I do not need to know everything. It is okay to master one thing and go deep into it and in tech you need to be adaptable, so focus on building that capability instead of just reacting instantly.

I am still in this phase:-

still exploring
still figuring things out.
still confused

but I have changed the way of doing things:-

less panic, focus more
more consistency
more patience

If you are also someone like me?

Feels like you are not progressing enough & Unsure about you path

You're not alone. This is uncomfortable but necessary.

Maybe this part of the journey is just about continuing -> learning, trying, failing, and still showing up.

And maybe one day, all of this the confusion, the switching, the doubt, the anxiety all will turn into something meaningful.