The Problem With Most Data Science Roadmaps
Most data science roadmaps online are either overwhelming lists of 50 technologies, or oversimplified "learn Python in a weekend" nonsense. Neither helps a real beginner. This roadmap is different — it tells you exactly what to learn, in what order, and gives you an honest estimate of how long each stage takes.
Stage 1 — Python Fundamentals (4-6 weeks)
Everything in data science runs on Python. Before you touch any data, you need to be comfortable with the basics: variables, data types, loops, functions, and basic file handling. You do not need to master Python before moving on — you just need to be comfortable enough to read and write simple scripts.
# What you should be able to write after Stage 1
def calculate_average(numbers):
total = sum(numbers)
return total / len(numbers)
scores = [85, 92, 78, 96, 88]
print(calculate_average(scores))
Stage 2 — Data Analysis with Pandas and NumPy (4-6 weeks)
This is where data science actually starts. Pandas is your main tool for working with tabular data — loading CSVs, cleaning columns, filtering rows, grouping data. NumPy handles numerical operations and is the foundation that Pandas is built on. Work through at least 3 real datasets during this stage. Kaggle has hundreds of free datasets you can download and practise on.
Stage 3 — Data Visualisation (2-3 weeks)
Data that you can't communicate clearly is useless. Matplotlib gives you control over every pixel of your charts. Seaborn gives you beautiful statistical plots in a few lines. Learn both. Focus especially on: line charts for trends, bar charts for comparisons, scatter plots for relationships, and heatmaps for correlations.
Stage 4 — Statistics and Probability (3-4 weeks)
This is the stage most beginners skip — and it's why they struggle later. You don't need a degree in statistics, but you do need to understand: mean, median, mode, standard deviation, distributions, correlation vs causation, hypothesis testing, and p-values. These concepts come up in every data science interview.
Stage 5 — Machine Learning with Scikit-learn (6-8 weeks)
Now you're ready for machine learning. Scikit-learn makes it surprisingly accessible. Start with supervised learning — linear regression for predicting numbers, logistic regression for predicting categories. Then move to decision trees and random forests. Learn how to evaluate models properly using train/test splits and cross-validation.
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
model = LinearRegression()
model.fit(X_train, y_train)
predictions = model.predict(X_test)
Stage 6 — Build a Portfolio (ongoing)
No portfolio means no job. Start building projects from Stage 2 onwards — don't wait until you feel "ready". Three solid projects on GitHub will get you more interviews than any certification. Each project should have a clear question it answers, clean code, a README that explains what you did, and visualisations of your findings.
How Long Does It Actually Take?
If you study consistently for 1-2 hours per day, you can complete this roadmap in 6-9 months. If you have more time — say 4-5 hours a day — you can compress it to 3-4 months. The biggest factor is not how smart you are. It's how consistent you are. Thirty minutes every day beats five hours on a Sunday.
Want to Move Faster?
The fastest way to learn data science is with direct feedback from someone who already knows it. When you get stuck on your own, you can spend hours going in circles. In a 1-on-1 session, that same problem gets resolved in minutes — and you understand why, not just how. That's why my students consistently move through this roadmap 2-3x faster than people learning alone.
Want Help Applying This?
Book a 1-on-1 session and we'll work through this topic together with your actual code and data.
Book a Session