Roadmaps › Data & AI
Data Analyst Roadmap
Turn raw data into decisions — Excel, SQL, Python, statistics and dashboards — one of the most in-demand entry-level roles.
Level: Beginner · Duration: 4–5 months
Step 1: Excel / Google Sheets
Still the most used data tool in companies.
Spreadsheets are still the most used data tool in companies. Many analyst interviews start with an Excel test, and quick spreadsheet skills save you hours every week.
- Formatting and organising data in tables
- Essential formulas and lookups
- Pivot tables and charts
- Cleaning messy data
Resources: ▶ Data Analyst Bootcamp
Formulas
SUMIFS, COUNTIFS, IF, XLOOKUP / VLOOKUP.
Formulas calculate and combine data automatically. A small set of functions covers most real work.
- SUM, AVERAGE, COUNT and their IF/IFS versions
- IF, AND, OR
- VLOOKUP / XLOOKUP and INDEX-MATCH
- Absolute vs relative references ($A$1)
Pivot tables & charts
Summarise thousands of rows in seconds.
Pivot tables summarise thousands of rows in seconds — totals by month, region or product — and charts make the result easy to understand.
- Creating and filtering pivot tables
- Grouping by dates
- Bar, line and pie charts — and when to use each
- Slicers
Cleaning data
Text functions, removing duplicates, data validation.
Real data is messy: extra spaces, duplicates, inconsistent spellings. Cleaning it is a big part of an analyst's job.
- TRIM, UPPER/PROPER, TEXT functions
- Removing duplicates
- Text to columns and splitting
- Data validation
Step 2: Statistics basics
Know what the numbers really say.
Statistics helps you describe data and decide whether a pattern is real or just chance. You do not need advanced maths — just a solid grasp of the basics.
- Measures of centre and spread
- Distributions
- Correlation
- Basic inference
Descriptive statistics
Mean, median, mode, variance, standard deviation.
Descriptive statistics summarise data with a few numbers: where the middle is and how spread out the values are.
- Mean, median and mode — and when the median is better
- Range, variance and standard deviation
- Percentiles and quartiles
- Box plots
Distributions
Normal distribution, percentiles, outliers.
A distribution shows how often each value occurs. The normal (bell) curve appears everywhere, and spotting outliers is an everyday analyst task.
- Histograms
- Normal distribution and z-scores
- Skewness
- Detecting outliers with IQR
Correlation vs causation
Why a relationship does not prove a cause.
Two things moving together does not mean one causes the other. Explaining this clearly protects companies from bad decisions.
- Correlation coefficient
- Scatter plots
- Confounding variables
- Why experiments (A/B tests) show causation
Hypothesis testing (optional)
p-values, A/B tests.
Hypothesis testing checks whether a difference — say, between two versions of a web page — is real or due to chance.
- Null and alternative hypotheses
- p-values and significance
- A/B testing basics
Step 3: SQL
Get data out of databases yourself.
SQL is the most important technical skill for a data analyst: it lets you pull exactly the data you need from company databases yourself.
- Querying and filtering
- Aggregating
- Joining tables
- Advanced analytics with window functions
Resources: SQL practice (live database) · SQL documentation · SQL cheat sheet · PostgreSQL documentation · ▶ SQL Tutorial in Hindi · ▶ SQL Complete Course (MySQL)
SELECT, WHERE, ORDER BY
Filter and sort.
Every query starts by choosing columns, filtering rows and sorting the result.
- SELECT specific columns and aliases
- WHERE with AND/OR, IN, BETWEEN, LIKE
- ORDER BY and LIMIT
- Handling NULL values
GROUP BY & aggregates
COUNT, SUM, AVG with HAVING.
Aggregates turn many rows into summary numbers — revenue per month, orders per customer.
- COUNT, SUM, AVG, MIN, MAX
- GROUP BY one or more columns
- HAVING vs WHERE
- COUNT(DISTINCT …)
Joins
INNER, LEFT, self joins.
Business data is spread across tables (customers, orders, products). Joins combine them.
- INNER JOIN
- LEFT JOIN and finding missing matches
- Joining more than two tables
- Self joins
Resources: Notes: SQL Essentials and Joins
Window functions
RANK, ROW_NUMBER, running totals.
Window functions calculate across related rows without collapsing them — rankings, running totals, month-over-month change. They are favourites in analyst interviews.
- ROW_NUMBER, RANK, DENSE_RANK
- PARTITION BY
- Running totals with SUM() OVER
- LAG and LEAD
Resources: Practise: salary rank in department · Practise: running revenue
CTEs & subqueries (optional)
Break big queries into steps.
CTEs (WITH clauses) break big queries into readable steps; subqueries nest one query inside another.
- Writing CTEs
- Subqueries in WHERE and FROM
- Choosing readability over cleverness
Step 4: Python for data
Automate analysis and handle bigger data.
Python handles data that is too big or too complex for spreadsheets and lets you repeat an analysis with one click.
- Python basics for analysis
- NumPy and Pandas
- Visualisation libraries
- Notebooks
Resources: ▶ Python for Beginners · Python documentation
Python basics
Enough Python: lists, dicts, functions, loops.
You only need core Python — variables, lists, dictionaries, loops and functions — before moving to data libraries.
- Lists, dicts and loops
- Functions
- Reading files
Resources: Python roadmap
NumPy
Fast arrays and maths.
NumPy provides fast arrays and vectorised maths — the foundation Pandas is built on.
- Creating arrays
- Vectorised operations instead of loops
- Basic statistics on arrays
Pandas
Read CSV/Excel, clean, filter, group and merge.
Pandas is the main tool for analysis in Python: load data into DataFrames, clean it, filter it, group it and merge it.
- Reading CSV and Excel into DataFrames
- Selecting, filtering and sorting
- groupby and aggregations
- merge and concat
- Handling dates
Matplotlib / Seaborn
Charts from code.
Charts reveal patterns that tables hide. Matplotlib gives full control; Seaborn makes attractive statistical charts quickly.
- Line, bar, histogram and scatter plots
- Titles, labels and readable charts
- Seaborn for distributions and heatmaps
Jupyter notebooks (optional)
Mix code, charts and notes.
Notebooks mix code, results, charts and notes in one document — ideal for exploring data and sharing an analysis.
- Running cells
- Markdown cells for explanations
- Sharing notebooks on GitHub or Kaggle
Step 5: Data cleaning & EDA
Most of the job is cleaning data.
Most of an analyst's time goes into cleaning data and exploring it. Good cleaning prevents wrong conclusions; good exploration finds the interesting questions.
- Spotting and fixing data quality issues
- Exploratory data analysis
- Documenting what you changed
Missing values & duplicates
Find and fix them.
Missing and duplicate records distort results. Decide case by case whether to drop, fill or flag them.
- Finding nulls and duplicates
- Dropping vs filling (mean, median, forward-fill)
- Fixing inconsistent categories and data types
Exploratory data analysis
Ask questions, plot distributions, spot trends.
EDA means asking questions of the data: how is each column distributed, what changes over time, which groups differ?
- Summary statistics per column
- Distributions and outliers
- Trends over time
- Comparing groups
Step 6: Dashboards & BI
Share insights with non-technical people.
Business intelligence tools turn analysis into dashboards that managers check every day. Power BI and Tableau skills appear in most analyst job descriptions.
- Connecting data sources
- Building interactive visuals
- Designing clear dashboards
- Publishing and sharing
Resources: ▶ Data Analyst Bootcamp
Power BI or Tableau
Connect data, build visuals, publish.
Pick one tool — Power BI is very common in Indian companies — and learn to go from raw data to a published dashboard.
- Importing and transforming data (Power Query)
- Building charts, tables and cards
- Filters and slicers
- Publishing reports
Dashboard design
One question per chart; highlight what matters.
A good dashboard answers specific questions at a glance. Less is more: highlight what matters and remove clutter.
- One key question per chart
- Choosing the right chart type
- Layout, colour and consistency
- Designing for the audience
DAX basics (optional)
Calculated measures in Power BI.
DAX is Power BI's formula language for calculated measures like year-over-year growth.
- Measures vs calculated columns
- CALCULATE and filters
- Time intelligence functions
Step 7: Business thinking
Analysts are paid for answers, not charts.
Companies hire analysts for answers, not charts. Understanding the business — what drives revenue, cost and customer behaviour — makes your analysis useful.
- Understanding business goals
- Choosing the right metrics
- Communicating recommendations
KPIs & metrics
Revenue, retention, conversion, churn.
KPIs are the numbers a business watches to know if it is doing well. Know the common ones and how they are calculated.
- Revenue, average order value, conversion rate
- Retention and churn
- Customer acquisition cost
- Choosing a metric for a business question
Storytelling
Problem → analysis → recommendation.
Data storytelling turns analysis into action: what is the problem, what did the data show, and what should we do?
- Problem → insight → recommendation
- Writing a short summary for managers
- Presenting findings clearly
Step 8: Portfolio projects
Show real analysis on real data.
A portfolio shows you can analyse real data end to end. Use public datasets and explain your findings clearly.
- Choosing datasets and questions
- Documenting your process
- Sharing on GitHub, Kaggle or a portfolio site
Resources: Kaggle datasets
SQL case study
e.g. analyse an e-commerce orders database.
Answer business questions from a realistic database — for example, an e-commerce store's orders and customers.
- Top products and customers
- Monthly revenue and growth
- Customer retention analysis
Python EDA project
A notebook with clear findings.
Take a public dataset, clean it, explore it and write up the three or four most interesting findings with charts.
- Data cleaning steps
- Clear charts and explanations
- A conclusions section
Dashboard project
A published Power BI / Tableau dashboard.
Build and publish an interactive dashboard, such as sales performance by region and month.
- Defining the questions it answers
- Interactive filters
- Publishing and sharing the link
Step 9: Next steps: Data Science (optional)
Where to go after data analysis.
Once you are comfortable with analysis, data science adds predictive modelling with machine learning.
- Machine learning basics
- Deeper statistics
- Model evaluation
Resources: ▶ 100 Days of Machine Learning
Machine learning basics
scikit-learn: regression, classification.
Machine learning learns patterns from data to make predictions. scikit-learn makes the common models easy to try.
- Regression and classification
- Train/test split
- Evaluating models (accuracy, precision, recall)
Advanced statistics
Probability and inference.
Deeper statistics helps you understand models and design reliable experiments.
- Probability distributions
- Confidence intervals
- Regression analysis