■ Data Science

Complete Roadmap

Python · SQL · Machine Learning · Power BI · FastAPI · Docker · Cloud

Modules Weeks Topics Projects

12 30 50+ 3+

■ 30-Week Master Plan

Week Focus

Week 1–4 Python Foundations + Git basics

Week 5–8 SQL mastery + Database design

Week 9–11 EDA with Pandas + Visualization

Week 12–14 Power BI Dashboard + DAX

Week 15–19 Machine Learning core + Projects

Week 20–21 Statistics Applied

Week 22–23 FastAPI Backend + Deployment

Week 24–25 Docker + Cloud + CI/CD

Week 26–28 End-to-End Projects (polish)

Week 29–30 Resume + LinkedIn + Interview Prep

 01 ■ Programming Foundations (Python)

Python is the #1 language for Data Science. Isko achhe se seekhna zaroori hai kyunki baaki sab iske upar
build hoga.

■ Python Syntax & Control Flow

• Variables, data types (int, str, list, dict, tuple, set)
• if/elif/else conditions, for/while loops
• List comprehensions — power feature of Python
• Practice: HackerRank Python Easy problems (20 problems minimum)

■ Functions & Modules

• Define functions: def, return, *args, **kwargs
• Lambda functions for short operations
• Import modules: math, os, sys, datetime
• Create your own module (.py file import karna)

■ OOP — Object Oriented Programming

• Classes and objects — blueprint concept
• Constructor (__init__), self keyword
• Inheritance: parent class se child class banana
• Encapsulation aur polymorphism basics

■ Error Handling & Logging

• try / except / finally blocks
• Custom exceptions banana
• logging module — print() se better practice
• Debug karna: breakpoints aur stack traces padhna

■ Virtual Environments
• venv: python -m venv myenv | activate karna
• conda: conda create, activate, install
• [Link] banana (pip freeze > [Link])
• Har project ke liye alag environment — professional habit

■ Code Quality
• PEP8 style guide: naming, spacing, line length
• flake8 / pylint se linting karna
• Black formatter — auto-format tool
• Docstrings likhna (functions ko document karna)

■ Timeline: 3–4 weeks | Tools: VS Code + Python 3.10+

 02 ■■ SQL & Databases

SQL Data Analyst/Scientist ki sabse important skill hai. 80% interviews mein SQL hoga. Isko master karo.

■ Basic Queries
• SELECT, FROM, WHERE — data filter karna
• ORDER BY (ASC/DESC), LIMIT, DISTINCT
• Aliases: AS keyword ka use
• Practice: SQLZoo, LeetCode SQL Easy problems

■ JOINs — Most Important!

• INNER JOIN — dono tables mein match data
• LEFT JOIN — left table ka saara data + match
• RIGHT JOIN — right table ka saara data
• Self JOIN aur multiple table joins

■ Aggregations
• GROUP BY — data groups mein todna
• Aggregate functions: COUNT, SUM, AVG, MIN, MAX
• HAVING — grouped data pe filter (WHERE nahi chalta yahan)
• Nested aggregations handle karna

■ Subqueries & CTEs

• Subqueries: SELECT ke andar SELECT
• WITH clause (CTE) — Common Table Expressions
• CTEs se complex queries readable banti hain
• Correlated subqueries — advanced concept

■ Window Functions
• ROW_NUMBER() — har row ko number dena
• RANK() aur DENSE_RANK() — ranking with ties
• LAG() / LEAD() — pichla/agla row value lena
• OVER(PARTITION BY ... ORDER BY ...) syntax

■ Database Design & Optimization

• Normalization: 1NF, 2NF, 3NF — redundancy hatana
• Primary Key, Foreign Key, Constraints
• Indexes — query speed badhane ke liye
• EXPLAIN / EXPLAIN ANALYZE se query plan dekhna

■ Timeline: 3–4 weeks | Tools: MySQL Workbench ya PostgreSQL (pgAdmin)

 03 ■ Data Analysis (EDA)

Exploratory Data Analysis — koi bhi ML model banane se pehle data ko samajhna zaroori hai. Pandas
tumhara best friend hai.

■ Pandas — Core Library

• DataFrame aur Series — pandas ke basic units
• read_csv(), head(), info(), describe()
• merge() — SQL JOIN jaisa kaam pandas mein
• groupby() — split-apply-combine pattern
• pivot_table() — data summarization

■ Data Cleaning
• Missing values: isnull(), fillna(), dropna()
• Duplicate rows: duplicated(), drop_duplicates()
• Data type correction: astype(), pd.to_datetime()
• String cleaning: [Link](), [Link](), [Link]()

■ Outlier Detection
• IQR method: Q1 - 1.5*IQR to Q3 + 1.5*IQR
• Z-score method: |z| > 3 are outliers
• Box plots se visual detection
• Domain knowledge se decide karo: remove ya cap?

■ Feature Understanding
• Univariate analysis: ek variable ka distribution
• Bivariate analysis: do variables ka relation
• Correlation matrix: heatmap se relationships
• Feature transformation: log, sqrt, binning

■ Visualization
• Matplotlib: [Link](), [Link](), [Link]()
• Seaborn: [Link](), [Link](), [Link]()
• Always add: title, xlabel, ylabel, legend
• Plotly — interactive charts (bonus skill)

■ Timeline: 3 weeks | Dataset: Kaggle Titanic, House Prices

04 ■ Data Visualization & BI

Data ko story mein convert karo. Business stakeholders code nahi dekhte — wo dashboards dekhte hain.
■ Dashboard Design Principles
• KPIs clearly dikhao — most important numbers upar
• Storytelling: context → data → insight → action
• Color use karo purposefully — too many colors = confusion
• Mobile-friendly layouts — responsive design

■ Power BI / Tableau
• Power BI: Microsoft ecosystem, DAX language, free desktop
• Tableau: more visual flexibility, industry standard
• Recommendation: Power BI seekho (job demand zyada India mein)
• Filters, slicers, drill-down — interactive elements

■ DAX — Power BI
• Calculated columns vs measures — fark samajhna
• Basic measures: SUM(), COUNT(), AVERAGE(), DIVIDE()
• Time intelligence: TOTALYTD(), SAMEPERIODLASTYEAR()
• CALCULATE() — most powerful DAX function

■ Data Modeling
• Star Schema: fact table + dimension tables
• Snowflake schema — normalized dimensions
• Relationships: one-to-many, many-to-many
• Import vs DirectQuery mode

■ Timeline: 2–3 weeks | Goal: 1 live dashboard on Power BI Service

05 ■ Machine Learning (Core)

ML matlab data se automatically patterns seekhna. Scikit-learn library use karenge — industry standard
hai.

■ ML Workflow
• Problem define → Data collect → EDA → Feature Eng → Model → Evaluate → Deploy
• Train/test split: 80-20 ya 70-30
• Cross-validation: K-Fold (k=5 ya 10)
• Data leakage se bachna — sabse common mistake

■ Supervised — Regression
• Linear Regression: straight line fit, MSE/RMSE metrics
• Assumptions: linearity, normality, no multicollinearity
• Ridge (L2) aur Lasso (L1) regularization
• Use case: house price prediction, salary estimation

■ Supervised — Classification
 • Logistic Regression: probability output (0 to 1)
• Decision Trees: if-else tree structure, Gini impurity
• Random Forest: bagging + many trees = powerful
• KNN: k nearest neighbors se classify karna

■ Unsupervised Learning
• K-Means: data ko k clusters mein todna
• Elbow method: optimal k choose karna
• PCA: high dimensions ko reduce karna, variance preserve
• Use case: customer segmentation, anomaly detection

■ Evaluation Metrics
• Classification: Accuracy, Precision, Recall, F1-Score
• Confusion Matrix: TP, TN, FP, FN samajhna
• Regression: RMSE, MAE, R-squared
• ROC-AUC curve: threshold-independent performance

■ Feature Engineering
• Encoding: One-Hot (pd.get_dummies), Label Encoding
• Scaling: StandardScaler, MinMaxScaler
• Handling imbalanced data: SMOTE, class_weight
• Feature selection: correlation, feature importance

■ Timeline: 4–5 weeks | Library: scikit-learn, pandas, numpy

06 ■ Statistics (Applied)

ML aur Data Science ki neenv statistics hai. Concepts samajho — formulas ratto mat.

■ Descriptive Statistics
• Mean, Median, Mode — central tendency
• Variance aur Standard Deviation — spread of data
• Skewness aur Kurtosis — distribution shape
• Percentiles aur Quartiles (Q1, Q2, Q3)

■ Probability & Distributions

• Probability basics: P(A), P(A|B), Bayes theorem
• Normal Distribution: 68-95-99.7 rule
• Binomial Distribution: success/failure experiments
• Central Limit Theorem — statistics ka golden rule

■ Hypothesis Testing
• Null hypothesis (H0) vs Alternative (H1)
• p-value: agar p < 0.05 → reject H0
• t-test: do groups ka mean compare karna
 • Chi-square test: categorical variables ke liye
• Confidence Intervals: 95% CI ka matlab

■ Timeline: 2 weeks | Resource: StatQuest YouTube channel

07 ■■ Backend & APIs (Product Skills)

Tera ML model sirf Jupyter notebook mein nahi rehna chahiye. Use karo FastAPI se production-ready API
banana.

■ REST API Concepts

• HTTP methods: GET, POST, PUT, DELETE
• Status codes: 200 (OK), 404 (Not Found), 500 (Server Error)
• Request/Response structure: headers, body, params
• JSON format — APIs ki language

■ FastAPI (Recommended)
• FastAPI setup: pip install fastapi uvicorn
• Route define karna: @[Link]('/'), @[Link]('/predict')
• Pydantic models se request validation
• Async endpoints — high performance

■ Authentication Basics
• JWT tokens: login karo → token milta hai → API use karo
• Session-based auth vs Token-based auth
• API Keys — simple authentication
• OAuth2 basics (Google login jaisa)

■ API Documentation
• Swagger UI: FastAPI mein auto-generated (/docs route)
• Endpoint document karna: descriptions, examples
• Postman se API test karna
• Error messages meaningful banao

■ Timeline: 2 weeks | Goal: ML model ko /predict endpoint pe serve karo

08 ■ Deployment & DevOps Basics

Code likha to theek hai — duniya tak pahunchana seekho. Deployment se hi 'live project' banta hai.

■ Git & GitHub

 • git init, add, commit, push, pull — daily commands
• Branching: main branch pe directly mat push karo
• Pull Requests (PRs) — code review process
• GitHub README likhna — portfolio ka face

■ Docker Basics
• Container concept: environment ko box mein pack karna
• Dockerfile likhna — image build instructions
• docker build, docker run, docker-compose
• 'Works on my machine' problem Docker solve karta hai

■ Cloud Deployment
• Render / Railway: simplest free deployment (start yahan se)
• AWS EC2: virtual machine pe app deploy karna
• AWS S3: static files, models, datasets store karna
• Environment variables: secrets code mein mat likho

■ CI/CD & Monitoring

• GitHub Actions: code push hote hi auto-test/deploy
• Basic CI pipeline: test → build → deploy
• Logging: error track karna production mein
• Uptime monitoring: kya app live hai?

■ Timeline: 2 weeks | Goal: App deploy karo aur live URL share karo

09 ■ End-to-End Projects (Portfolio)

Sabse important section. Companies tumhara code dekhti hain, certificates nahi. 2-3 strong projects = job.

■ Project Structure
• Business use-case: real problem define karo
• Data collection: Kaggle, web scraping, APIs, simulation
• EDA + Feature Engineering → Model Building → Evaluation
• API serving: model ko endpoint pe lagao

■ Project Ideas (Do These!)

• Project 1: Customer Churn Prediction (ML + FastAPI + Dashboard)
• Project 2: Sales Forecasting (Time Series + Power BI)
• Project 3: Sentiment Analysis (NLP + deployment on Render)
• Bonus: LifeOS jaisi app — full stack + ML integration

■ GitHub Presentation
• README mein: problem, approach, results, screenshots
• Architecture diagram add karo
• Demo video ya live link mandatory
 • Clean code, proper folder structure

■ Dashboard Integration
• Model results ko dashboard se connect karo
• Real-time predictions dikhao
• Streamlit — Python se quick dashboard banana
• Power BI report embed karna web app mein

■ Timeline: Ongoing — har module ke saath ek mini project karo

10 ■ System Thinking

Senior roles mein system design bahut important hai. Data apps kaise scale hote hain — ye samajhna
chahiye.

■ Data Pipelines (ETL)

• Extract: source se data lena (API, DB, files)
• Transform: clean, aggregate, enrich karna
• Load: destination pe save karna (Data Warehouse)
• Tools: Apache Airflow, dbt, Pandas pipelines

■ Processing Paradigms
• Batch processing: scheduled, large data, daily reports
• Real-time / Streaming: Kafka, Spark Streaming
• Lambda architecture: batch + streaming combined
• When to use which? — it depends on latency requirement

■ Scalability Basics
• Caching: Redis se frequently accessed data cache karo
• Load balancing: traffic distribute karna
• Horizontal vs Vertical scaling
• Database sharding aur replication basics

■ Timeline: 1–2 weeks | Resource: System Design Primer (GitHub)

11 ■ Resume & Profile

Tera resume = tera sales pitch. ATS systems pehle screen karte hain, phir human. Dono ke liye optimize
karo.

■ Resume Tips
 • Impact-based bullets: 'Reduced churn by 15% using XGBoost model'
• Quantify everything: %, time saved, records processed
• ATS-friendly: simple format, no tables in header, standard fonts
• 1 page (fresher) ya 2 pages max — recruiter ke paas time nahi

■ GitHub Profile
• Pinned repos: sirf best 6 projects pin karo
• [Link]: har repo mein achha README mandatory
• Contribution graph green rakho — consistency dikhao
• Profile README: ek intro file banao ([Link]/)

■ LinkedIn Optimization
• Headline: 'Data Scientist | Python | SQL | ML | Power BI'
• About: 3-4 lines mein skills + projects + goal
• Projects section: screenshots aur live links add karo
• Recommendations: seniors/professors se manga

■ Timeline: 1 week | Tool: Overleaf (LaTeX resume) ya Canva ATS template

12 ■ Interview Preparation

Preparation structured honi chahiye. Random padhai se nahi hoga — targeted practice karo.

■ SQL Interview Questions

• Practice: top 50 LeetCode SQL problems
• Common: second highest salary, running totals, rank without RANK()
• Case studies: revenue trends, cohort analysis
• Time yourself: 20 min per medium problem

■ ML Concept Questions
• Explain: bias-variance tradeoff, overfitting solutions
• When to use which algorithm? — practical knowledge
• Feature selection techniques explain karo
• Cross-validation types aur use cases

■ Case Studies
• Business problem: 'Revenue girti ja rahi hai — kya karoge?'
• Structure: Understand → Hypothesize → Analyze → Recommend
• Always ask clarifying questions pehle
• Data availability assumptions clearly bolna

■ Project Deep Dive

• STAR method: Situation, Task, Action, Result
• Har project mein: problem, approach, challenges, impact
• Why this algorithm? Why not another?
 • What would you improve? — shows growth mindset

■ Behavioral Questions
• 'Tell me about yourself' — 2 min structured pitch ready rakho
• 'Biggest failure?' — honest + what you learned
• 'Why Data Science?' — genuine answer, not cliche
• Research the company: products, data use, recent news

■ Timeline: 2–4 weeks before interviews | Mock interviews karo!

 ■ Final Target — Job-Ready Definition

■ 2–3 strong end-to-end projects deployed with live URLs

■ SQL strong — LeetCode Medium level solve kar sako

■ 1 live dashboard on Power BI Service ya Streamlit Cloud

■ 1 ML model deployed as REST API (FastAPI on Render)

■ Resume ATS-friendly + GitHub profile clean & polished

"Consistency beats intensity. Roz thoda — 30 weeks mein job ready!"