Data Science Roadmap for Beginners

Data science is one of the most sought-after careers in India right now.

Companies across banking, e-commerce, healthcare, logistics, and technology are hiring professionals who can extract meaningful insight from data and paying very well to do so.

Yet many beginners are paralysed by the breadth of skills the field requires.

This roadmap gives you a clear, step-by-step path from complete beginner to job-ready, tells you what to learn in what order and why, and points you to the best resources at each stage.

Follow it consistently and you will be ready to apply for entry-level data roles in 9 to 12 months.

What does a data scientist actually do?

Data science is not a single job -- it is an umbrella for roles with different emphases.

Data analysts focus on querying data, creating dashboards, and communicating insights using SQL, Excel, and BI tools.

Data scientists build predictive models using statistics and machine learning.

ML engineers deploy models.

Data engineers build the infrastructure that makes data available.

Most beginners should aim for a data analyst role first.

The barrier to entry is lower, the skills are learned faster, and it gives you invaluable context about business problems and data quality that makes you a much better data scientist when you step into that role.

The learning path for both overlaps significantly in the early stages.

Stage 1: Mathematics and statistics

You do not need a mathematics degree, but you do need a working understanding of the mathematical concepts that underpin the algorithms you will use.

The core areas: descriptive statistics, probability, and linear algebra basics.

You will use these constantly when interpreting model results and understanding why a model performs well or poorly.

Descriptive statistics: mean, median, mode, standard deviation, variance, percentiles, and distributions.

Probability: conditional probability, Bayes theorem, and probability distributions.

Linear algebra: vectors, matrices, and matrix multiplication at an intuitive level.

Hypothesis testing: p-values, confidence intervals, and A/B testing.

Khan Academy and StatQuest with Josh Starmer on YouTube are exceptional free resources.

Spend 4 to 6 weeks here, focusing on understanding rather than memorisation.

Descriptive statistics: mean, median, mode, standard deviation, percentiles
Probability: distributions, conditional probability, Bayes theorem
Linear algebra: vectors, matrices, eigenvalues conceptually
Hypothesis testing: p-values, confidence intervals, A/B testing

Stage 2: Python for data science

Python is the language of data science.

Its ecosystem of libraries -- NumPy, Pandas, Matplotlib, and Scikit-learn -- makes it the most productive environment for every stage of the data workflow.

Start with Python basics: variables, loops, functions, lists, and dictionaries.

Then move into the data science libraries.

Pandas is the workhorse of practical data work.

It handles loading, cleaning, transforming, filtering, and aggregating data.

NumPy provides the numerical computing foundation.

Matplotlib and Seaborn handle visualisation.

Spend serious time on Pandas specifically -- it is used daily in every data role.

Resources: official Pandas documentation, Kaggle free Python and Pandas courses, and Google Colab for free browser-based Python.

Stage 3: SQL and data wrangling

SQL is the language of data.

Even if you spend most of your career in Python, you will use SQL every day to query databases, join tables, aggregate results, and extract data.

It is also the most commonly tested skill in data analyst and scientist interviews across India.

Core SQL skills: SELECT, FROM, WHERE, GROUP BY, HAVING, ORDER BY, JOINs, subqueries, window functions, and CTEs.

Window functions -- ROW_NUMBER, RANK, LAG, LEAD, and running totals -- separate intermediate SQL users from beginners.

Practice on LeetCode SQL section, Mode Analytics SQL Tutorial, and HackerRank SQL challenges.

Stage 4: Exploratory data analysis

Exploratory data analysis (EDA) is the practice of investigating a dataset to understand its structure, identify patterns, find anomalies, and generate hypotheses before modelling.

Many beginners rush to build models without understanding the data, which leads to misleading results.

EDA includes: checking for missing values, understanding distributions using histograms and box plots, examining correlations between variables, identifying outliers, and creating clear visualisations.

Seaborn and Matplotlib cover most EDA needs.

Plotly and Streamlit allow interactive dashboards.

Tableau and Power BI are important for analyst roles at companies that use BI tools.

Stage 5: Machine learning fundamentals

Machine learning is often seen as the glamorous heart of data science, but the bulk of a data scientist time is typically spent on data cleaning, feature engineering, and interpretation.

That said, understanding the core algorithms, when to apply them, and how to evaluate them is essential.

Start with Scikit-learn.

Learn the key algorithms: linear regression, logistic regression, decision trees, random forests, gradient boosting (XGBoost/LightGBM), k-means clustering, and k-nearest neighbours.

Critical evaluation concepts: train/test split, cross-validation, precision, recall, F1 score, ROC-AUC, and RMSE.

Andrew Ng Machine Learning Specialisation on Coursera is the single best structured course for this stage.

Stage 6: Real projects and Kaggle

No amount of learning substitutes for building real things.

Kaggle is the best platform for beginners.

Start with the Titanic survival prediction competition, then attempt two or three other competitions in areas you are interested in.

Kaggle Expert or Master ranking is a credible signal on a CV.

Good original project ideas: analysing a dataset from your field of interest (sports, finance, healthcare), building a price prediction model using real estate data, or creating a dashboard that visualises publicly available Indian government data.

A well-documented, original project tells a better story than any coursework completion certificate.

Making yourself job-ready

Once you have the technical foundations and a portfolio of real projects, the job search begins.

Tailor your CV to data roles specifically: lead with quantified achievements and specific technical skills.

Target data analyst roles at product companies, fintechs, e-commerce platforms, and healthcare companies rather than large consulting firms for your first role.

Interview preparation: practise SQL on LeetCode and HackerRank, be prepared to walk through an EDA on a new dataset in Python, review basic statistics and probability, and prepare to explain every project on your CV as if presenting to a non-technical audience.

Communication and storytelling are underrated interview skills for data roles -- the ability to translate a model output into a business recommendation is what makes a data professional truly valuable.

Frequently Asked Questions

No. A strong working knowledge of statistics is necessary but fully learnable through self-study. Many successful data scientists come from engineering, economics, or other quantitative backgrounds.

With consistent effort of 8 to 10 hours per week, most people are ready to apply for junior data analyst roles in 6 to 8 months. Full data scientist roles typically require 12 to 18 months of learning plus real project experience.

Python. It has overtaken R comprehensively in industry usage, has a far larger job market, and is more versatile beyond data science. R remains strong in academic research, but for India job market, Python is the clear choice.

Entry-level data analysts earn Rs 4 to 10 lakh per year. Junior data scientists with 1 to 3 years earn Rs 8 to 20 lakh. Senior data scientists with 5+ years earn Rs 20 to 50 lakh and beyond. Salaries are highest at product companies and fintechs.

Kaggle is a strong signal but not essential. Two or three original, well-documented projects on GitHub plus a Kaggle competition notebook are a stronger overall portfolio than only a high Kaggle rank.