0% found this document useful (0 votes)
11 views9 pages

Data Science Work Process

The document outlines the key responsibilities and processes involved in data science, emphasizing the importance of understanding business problems before data collection. It details the stages of data cleaning, feature engineering, model building, testing, and deployment, highlighting that data cleaning takes the majority of time. Ultimately, it stresses that a data scientist's role is more about problem-solving than just machine learning.

Uploaded by

bitubhai68
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
11 views9 pages

Data Science Work Process

The document outlines the key responsibilities and processes involved in data science, emphasizing the importance of understanding business problems before data collection. It details the stages of data cleaning, feature engineering, model building, testing, and deployment, highlighting that data cleaning takes the majority of time. Ultimately, it stresses that a data scientist's role is more about problem-solving than just machine learning.

Uploaded by

bitubhai68
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

PySolanki

Data
Science
Presented by
Page 1

Abhishek Next Slide


PySolanki Business
Understanding
What it really means:
Before touching data or code, a data scientist first
understands the problem.
In simple words:
👉 “What exactly are we trying to solve, and why?”
Real-life example:
A company says:
“Our customers are leaving.”
A data scientist asks:
What is “leaving”? (not buying again? uninstalling app?)
Who is the customer?
When do they leave?
How will the business use the result?
📌 Without business understanding, data science is useless.
Tools used:
Meetings
Excel / Docs
Page 2

Basic SQL to look at data Next Slide


(No fancy ML here)
PySolanki Data
Collection
What it really means:
Getting data from different places.
Simple explanation:
👉 “Data scientist collects raw data like collecting
ingredients before cooking.”
Where data comes from:
Company databases
Websites
APIs
CSV / Excel files
Technologies used:
SQL → company databases
APIs→ data from apps & services
BeautifulSoup → simple website scraping
Selenium → websites with buttons & login
Scrapy → large-scale web scraping
Page 3

Example:
Collecting product reviews from Amazon or Flipkart. Next Slide
PySolanki Data (THIS IS 60–70% OF THE JOB)

Cleaning
What it really means:
Fixing dirty, messy, broken data.
Simple explanation:
👉 “Data cleaning is like washing vegetables before
cooking.”
Common problems:
Missing values
Wrong data types
Duplicate rows
Spelling mistakes
Outliers (crazy values)
Example:
Age column has:
25, 30, NaN, 500, twenty
Tools used:
Python
Pandas
Page 4

NumPy Next Slide


📌 Most beginners think data science = ML
❌ Reality: data cleaning takes most time.
PySolanki Feature
Engineering
What it really means:
Creating useful information from raw data.
Simple explanation:
👉 “Turning raw data into smart data.”
Example:
You have: Input: 23/07/1998

Date of birth create Age Output columns:
Timestamp → create Hour, Day, Weekend Day → 23
Text → count positive/negative words Month → 7
Why it matters: Year→ 1998
Good features = better model
Bad features = useless model
Tools used:
Pandas
NumPy
Scikit-learn
Page 5

Basic math & logic


Next Slide
PySolanki Model
Building
What it really means:
Teaching a computer to find patterns and make predictions.
Simple explanation:
👉 “Showing the computer examples so it can learn.”
Example:
Input: customer data
Output: will customer leave or not?
Common models:
Linear Regression
Logistic Regression
Decision Trees
Random Forest
XGBoost
Tools used:
Scikit-learn
XGBoost
Page 6

TensorFlow / PyTorch (advanced)


📌 Models are important, but not everything. Next Slide
PySolanki Model
Testing & Optimisation
What it really means:
Checking if the model is actually good and improving it.

Simple explanation:
👉 “Testing before trusting.”

What data scientists do:


Test model on new data
Check accuracy, precision, recall
Tune parameters to improve performance

Example:
Model accuracy = 60% ❌
After tuning = 82%✅
Tools used:
Scikit-learn
Page 7

Cross-validation Next Slide


GridSearch / RandomSearch
PySolanki Model
Deployment
What it really means:
Putting the model into real-world use.

Simple explanation:
👉 “Making the model usable by real people.”

Example:
A website that predicts house prices
A system that flags fraud transactions

Technologies used:
Flask / FastAPI
Docker
AWS / GCP / Azure
CI/CD pipelines

📌 A model on your laptop = useless


Page 8

A deployed model = real data science Next Slide


PySolanki Final Truth
(VERY IMPORTANT)
❌ Data Scientist ≠ Only ML
✅ Data Scientist = Problem Solver
Real job breakdown:
20% Business understanding
30–40% Data cleaning
20% Feature engineering
10–15% Model building
Rest: testing & deployment
Page 9

End

You might also like