Data Science Roadmap for Beginners: 6 Months, Week by Week

McTaba runs coding and data bootcamps in Kenya. This guide is written to be useful whether or not you ever join one.

Imagine a Saturday morning. You have a laptop, a charger, a flask of tea and ten free hours. This plan fits into ten hours a week for 26 weeks. Two hours on four weekday evenings, two on Saturday.

Everything below is free. Python, SQL, pandas and Kaggle cost nothing to learn. You pay in hours, not shillings. If you want the wider picture of the job first, read our data science career map for Kenya.

The rule that holds the plan together: every month ends with something you built and put on GitHub. Watching videos without building is how people stay "learning" for two years.

Setup, day one

1. Open Google Colab. It runs Python in your browser, so a slow laptop is no excuse.
2. Make a free GitHub account and one repository called data-roadmap.
3. Make a free Kaggle account. Its notebooks and datasets save you downloads.

Month	Weeks	Focus	What you build
1	1 to 4	Python basics	A script that totals a made-up M-Pesa statement
2	5 to 9	SQL	Ten queries on a small shop database
3	10 to 13	pandas and charts	A notebook on a Kenyan public dataset
4	14 to 18	Statistics	A short report on an A/B test you simulate
5	19 to 22	First machine learning model	A model that predicts one thing and says how wrong it is
6	23 to 26	One finished project	A project with a README and a chart someone can read in a minute

If you would rather study with a class, a deadline and weekly feedback, the Data Science to AI Engineering programme covers the same ground in 5 months full time or 8 months part time. It starts on 11 January 2027 and costs KES 140,000. The free route below costs nothing but needs more discipline.

Weeks 1 to 4: Python

• Week 1: variables, numbers, strings. Use the official Python tutorial up to the chapter on data structures.
• Week 2: lists, dictionaries, loops. Write a loop that adds up ten made-up airtime top-ups.
• Week 3: functions and reading a CSV file with the csv module.
• Week 4: finish the script. Read a made-up statement file and print total in, total out and the largest single payment.

Our sibling guide, Python for data science: your first 30 days, lays out those four weeks in more detail.

Weeks 5 to 9: SQL

• Week 5: SELECT, WHERE, ORDER BY. Practise on SQLBolt, which runs in the browser.
• Week 6: GROUP BY and COUNT, SUM, AVG.
• Week 7: JOIN across two tables, such as customers and sales.
• Week 8: dates, CASE and subqueries.
• Week 9: write ten queries that answer questions a shop owner would ask, then save them in GitHub.

The queries you will use every week are in SQL for data analysis.

Weeks 10 to 13: pandas and charts

• Week 10: load a CSV, look at the first rows, count missing cells. Use the pandas getting started tutorials.
• Week 11: filter, sort, group. Answer "which day sells most?"
• Week 12: clean messy data. Phone numbers with and without 254, dates in two formats, names in capitals.
• Week 13: three charts with matplotlib. Month-three project: a notebook on a Kenyan public dataset. Where to find one is in Kenya open datasets.

Weeks 14 to 18: statistics

• Week 14: mean, median, spread. Use Khan Academy statistics and probability.
• Week 15: probability and the idea of a sample.
• Week 16: distributions, and why one rich customer wrecks an average.
• Week 17: confidence intervals and A/B tests.
• Week 18: write up a simulated A/B test in plain words, one page.

Read the statistics you actually need for data science alongside these weeks. It keeps you from drowning in theory.

Weeks 19 to 22: machine learning

• Week 19: the idea of training and test data. Use Kaggle Learn: Intro to Machine Learning.
• Week 20: a linear model that predicts a number, such as a day's sales.
• Week 21: a classifier that predicts yes or no, such as "will this loan be repaid?"
• Week 22: measure the mistakes. Say how often the model is wrong and what that costs.

Weeks 23 to 26: one finished project

• Week 23: pick one question a real business would ask. Write it in one sentence.
• Week 24: gather and clean the data.
• Week 25: analyse, then build one chart that answers the question.
• Week 26: write the README, tidy the notebook and show it to one person who does not code.

Choose the question first and the method second. A clean answer with a simple average beats a cloudy answer with a neural network.

Where this plan stops

Six months of free study gets you to a first portfolio piece. It does not give you a job, and no school can promise one. Production skills such as putting a model behind an API come after this roadmap.

Self-study stalls in three places: week 5 when SQL stops being obvious, week 17 when statistics gets abstract, and week 20 when nothing runs. If you stall for more than a week, ask for help, in a forum or a study group.

If you stall because life is busy, a class with fixed hours can fix that. The Data Science to AI Engineering programme runs four sessions a week, 7 PM to 9 PM for the 8-month part-time route. No coding, data or maths background needed.