Data & Analytics · Beginner

Data Analysis with Python

Pandas, NumPy and the unglamorous work of getting real data into a state worth analysing.

Lessons
18 lessons
Total runtime
22 hours of video
Modules
6 modules
Optional lessons
15 optional
Start course

Free. No account.

About this course

Most of data analysis is not analysis. It is loading a file that turns out to be malformed, discovering three spellings of the same city, deciding what to do about the eleven percent of rows with a missing value, and reshaping the whole thing into the layout your question actually needs. Courses that skip to the charts are teaching the last ten percent of the job.

This one starts with pandas, because pandas is where that work happens. The DataFrame is the central object — a table you can filter, group, join and reshape in a line or two — and getting genuinely fluent with it is the difference between analysis taking an hour and taking a day.

Corey Schafer's series is the careful teacher here and the right place to build habits. The freeCodeCamp course covers the same ground plus NumPy and the plotting libraries, and Keith Galli's single video is a good compressed alternative if you would rather see the whole workflow at once than in pieces.

One warning worth stating plainly: it is very easy to produce a confident-looking number from data that does not support it. Take the Statistics course alongside this. Knowing how to compute a correlation in pandas and not knowing what a correlation licenses you to claim is a genuinely dangerous combination. About six hours required, and around twenty-two if you take the optional lessons — a second and third pass over pandas, more on visualisation, and two complete projects on real datasets.

What you'll be able to do

  • Load messy real-world data from CSV, Excel and JSON and get it into a DataFrame
  • Filter, sort, group and aggregate data fluently
  • Handle missing and inconsistent values deliberately rather than by dropping rows
  • Join datasets together and reshape between wide and long layouts
  • Produce clear charts with matplotlib and seaborn
  • Know the limits of what your data can support you claiming

Curriculum

18 lessons · 22 hours

The DataFrame

5 lessons · 8h 30m

  1. 01CompletedPython Pandas Tutorial: Getting Started with Data AnalysisCorey SchaferVideoInstallation, loading data, and the DataFrame itself. Corey teaches slowly and correctly, which is what you want for the object every later lesson is built on.30m
  2. 02CompletedComplete Python Pandas Data Science TutorialKeith GalliVideoThe whole pandas workflow in one sitting — reading files, filtering, grouping and writing results back out. A good compressed complement to the slower series.1h 5m
  3. 03CompletedLearning Pandas for Data Analysis? Start HereOptionalRob MullaVideoTwenty minutes from a working data scientist on the handful of pandas operations that make up most real analysis. A good sanity check that you are learning the useful parts rather than the exhaustive ones.23m
  4. 04CompletedComplete Python Pandas Data Science Tutorial (Updated Edition)OptionalKeith GalliVideoA rerecording of the lesson above with current pandas syntax. Worth preferring if you hit deprecation warnings while following along with the older one.1h 34m
  5. 05CompletedPandas & Python for Data Analysis by ExampleOptionalfreeCodeCamp.orgVideoFive hours working through real datasets rather than demonstrating methods in isolation. The best long option here, and the closest to what the work actually feels like.4h 58m

The Arrays Underneath

2 lessons · 1h 22m

  1. 06CompletedPython NumPy Tutorial for BeginnersOptionalfreeCodeCamp.orgVideoEvery pandas column is a NumPy array underneath. An hour here explains why vectorised operations are fast and why looping over a DataFrame row by row is the slow beginner mistake.58m
  2. 07CompletedUltimate Guide to NumPy ArraysOptionalPython SimplifiedVideoShapes, axes and broadcasting specifically — the three things that produce confusing errors long after you can otherwise use the library.24m

Cleaning and Exploring

4 lessons · 4h 31m

  1. 08CompletedData Cleaning in PandasOptionalAlex The AnalystVideoDuplicates, missing values, inconsistent types and bad column names. This is genuinely most of the job, and it is the part tutorials skip by handing you a tidy CSV.39m
  2. 09CompletedExploratory Data Analysis in PandasOptionalAlex The AnalystVideoWhat to actually do when a dataset lands on you and nobody has told you what the question is. A repeatable first hour rather than a set of methods.32m
  3. 10CompletedExploratory Data Analysis with PandasOptionalRob MullaVideoThe same process on a different dataset, with more attention to the plots you make for yourself and throw away. Repetition is the point here.40m
  4. 11CompletedLearn Pandas in Under 3 HoursOptionalAlex The AnalystVideoFiltering, joins, indexing and cleaning in one continuous session. Useful as a consolidated reference once you know roughly what you are looking for.2h 40m

Showing the Result

4 lessons · 1h 52m

  1. 12CompletedLearn Matplotlib in 30 MinutesOptionalTech With TimVideoMatplotlib is ugly and universal, and every other Python plotting library is built on it. Half an hour so that when a chart needs adjusting you know which object to reach for.31m
  2. 13CompletedSeaborn Is The Easier MatplotlibOptionalNeuralNineVideoThe same plots with far less code, and defaults that are not embarrassing. Most exploratory charts should be made this way.23m
  3. 14CompletedSeaborn Crash CourseOptionalNeuralNineVideoThe fuller treatment — distributions, relationships and categorical plots, which is the vocabulary you need to choose a chart rather than default to a bar.36m
  4. 15CompletedIntro to Data Analysis and Visualization with PythonOptionalCS DojoVideoA short, complete arc from a raw file to a finished chart. Good for seeing how the pieces of this course fit together before you have all of them.22m

A Whole Project

2 lessons · 1h 41m

  1. 16CompletedPython Project for Data Analysis — Exploratory Data AnalysisOptionalRishabh MishraVideoOne dataset carried all the way through to conclusions, framed as something you could put in a portfolio. Do it with a different dataset afterwards; that second run is where the learning is.44m
  2. 17CompletedNew York Airbnb EDA ProjectOptionalZero AnalystVideoA messier, more realistic dataset than most tutorials use, with the cleaning left in rather than edited out.57m

The Full Toolkit

1 lesson · 4h 10m

  1. 18CompletedData Analysis with Python - Full Course for BeginnersfreeCodeCamp.orgVideoNumPy underneath, pandas on top, and matplotlib and seaborn for the output. The NumPy section is worth attention even though pandas hides most of it — array thinking is what makes vectorised code fast instead of a loop.4h 10m

Credits

Every lesson in this course was made by one of these channels. Tubeversity sequenced them; it did not make them. If a course is useful to you, the people below are the ones who earned it.