Python Data Analyst

Data work is mostly reshaping messy input into something a person can act on, and the two tools that never go away are Python and SQL. This route takes Python from first program to handling awkward strings and errors, spends real time on querying a database, then adds the algorithmic and machine-learning background that makes the rest legible. It stops where the library documentation starts, and says so.

6 stages · 32 steps

1. 1

   Python you can rely on

   Write a script that loads data, loops over it, and produces a number you would be willing to send to someone.

   • Language guidePythonThe language almost all data and AI tooling is written for.
   • LessonPython Variables and Data TypesNumbers, text and the difference between them, which is where a wrong total usually begins.
   • LessonPython For Loops and IterationEvery row-by-row calculation starts as one of these, even after a library hides it.
   • LessonPython ListsA column of values is a list long before any library turns it into a table.
   • LessonPython DictionariesGrouping and counting — the two operations analysis is made of — both live here.
   • LessonPython FunctionsWhere a one-off script turns into something you can rerun next month.
   • ExercisesPython exercisesEach exercise checks your code, so you find the gaps before a deadline does.
   • PlaygroundPython playgroundFor checking what one line does to one row before you run it over a million of them.
2. 2

   Real data is messy

   Clean inconsistent text, branch on conditions, and fail loudly instead of quietly producing a wrong number.

   • LessonPython StringsMost data cleaning is string work: trimming, splitting, normalising cases.
   • LessonPython ConditionalsDeciding what counts as valid, missing or an outlier is written as conditions.
   • Concept guideError HandlingA silent exception is how a report ends up confidently wrong.
   • ReferencePython cheat sheetFor the syntax you will look up weekly and never memorise.
   • Concept guidePython for AI and Machine LearningWhat the array and dataframe layer is for, before you read its own documentation.
3. 3

   Ask the database directly

   Answer a question with a query instead of exporting a file, and know which store you are querying.

   • Language guideSQLThe single most reusable skill in a data job.
   • PlaygroundSQL playgroundWrite queries against a small database in the browser; reading SQL teaches very little.
   • Concept guideSQL vs NoSQLSo you know why some data is not in a table at all.
   • Concept guidePostgreSQLIndexes and query plans are the difference between a two-second and a two-hour report.
   • Interview prepSQL interview questionsNearly every data interview has a SQL round; start reading it early.
4. 4

   Why the script is slow

   Reason about the cost of an operation, and pick a structure instead of waiting on a nested loop.

   • Concept guideBig O NotationThe vocabulary for explaining why a job that worked on a sample dies on the full dataset.
   • Concept guideHashing and Hash TablesDeduplication, joins and lookups all lean on this one idea.
   • Concept guideSorting AlgorithmsWorth knowing because sorting is often the expensive step you did not notice.
   • Practice problemGroup Word ClustersGrouping by a computed key, which is what a 'group by' does under the hood.
   • Practice problemBest Contiguous RunA single pass over a series — the shape of most time-series calculations.
5. 5

   From analysis towards models

   Describe honestly what a model is doing with your data, and where the numbers in an AI product come from.

   • AI courseAI FundamentalsPlain-English grounding first, so the terms in the next steps are not new at the same time.
   • Concept guideMachine Learning FundamentalsTraining, features and evaluation, in the language a data team uses.
   • Concept guideEmbeddings, ExplainedHow text becomes numbers you can cluster and compare.
   • Interview prepAI & Machine Learning interview questionsThe vocabulary check a data role's AI-adjacent round tends to apply.
6. 6

   Everyday tools and interview rounds

   Handle the formats and interview topics a data role actually puts in front of you.

   • Developer toolJSON Formatter & ValidatorMost API extracts arrive as JSON you have to read before you can parse it.
   • Developer toolUnix Timestamp ConverterFor the column of epoch seconds nobody documented.
   • Interview prepDatabase interview questionsWhere your tables came from, asked from the storage side rather than the query side.
   • Interview prepCoding & DSA interview questionsData roles ask lighter algorithm questions, but they do ask them.
   • Interview prepBehavioral interview questionsExpect to be asked about a time your analysis was wrong.

All roadmaps · Projects · Interview prep