Watch it happen. Then write it yourself.Learn data science by
Learn data science by
watching it happen.
Every method gets an animation that shows what actually moves — arrays stretching to broadcast, rows flying into groups, masks lighting up. Then you write the code yourself and real Python checks your answer.
This is what a lesson looks like
Broadcasting, from the NumPy track. Press play, or step through it yourself.
base0
1
2
0
1
2
3
0
0
0
0
10
10
10
10
20
20
20
20
shape (3, 4)
+
row0
1
2
3
1
2
3
4
shape (4,)
base is (3, 4). row is (4,) — just four values. Adding them looks like it shouldn't work: the shapes don't match.
The tracks
Each library gets the basics in depth, then a capstone that puts them together.
NumPy
28 lessonsArrays, shapes and broadcasting — the layer everything else is built on.
What an ndarray actually isMaking arraysdtypes, precision and castingMemory, strides and contiguity+240/104pandas
28 lessonsLabelled data: indexes, filtering, and split-apply-combine.
What a DataFrame actually isThe index, and alignmentdtypes in pandasGetting data in+240/109Matplotlib
20 lessonsFigures, axes and artists — what actually draws your chart.
Figure, Axes and Artistplt versus axYour first plotShowing and saving+160/62Machine Learning
35 lessonsThe core ideas, made visible — with scikit-learn.
What learning from data meansThe estimator APITrain and testYour first model+310/107
