Where to Go Next

Look back at everything this course taught you, from loading and cleaning data to charts, reports and tests, try a report across every week of the food truck's sales, plan an analysis of your own, and see where your data skills can go next.

Journeyman · Python · Updated

By

After this lesson you can

  • Choose the tool from this course that fits each step of an analysis of your own, from loading and cleaning to charting, sharing and testing

Key takeaways

  • You can now load data from CSV, Excel and JSON, clean it, answer questions by filtering, grouping, joining and reshaping it, chart the answers honestly, and share them as a report.
  • A report someone can trust has its logic in a module, a script that rebuilds it with one command, and tests that prove its numbers and charts.
  • The best next step is an analysis of your own, with its questions written down first, its cleaning in one function, its answers tested and its limits stated.
  • Each step of an analysis has a tool from this course, such as a cleaning function for a messy file and assert_frame_equal for a table you must trust.

When this course started, you added up a week of coffee orders by hand, one loop and one dictionary at a time. Look at what you do now: you load a file of thousands of rows, clean it, answer a manager's questions with a few lines each, draw the answer honestly, and prove the numbers with tests. The way to keep that, and to grow it, is to use it on data that matters to you.

Look how far you've come

After this course, you can:

  • set up a data workspace: a virtual environment with pandas and matplotlib, notebooks in VS Code you can trust because you run them top to bottom, and Git to get and hand in work;
  • load and describe data from CSV, Excel and JSON files: its size, columns and types, its gaps, its summary numbers and counts;
  • clean it honestly: missing values dropped or filled with a flag, types fixed, duplicates removed, text and dates tidied, impossible values caught, every step in one cleaning function;
  • answer questions: filter, sort, group and summarize, make pivot tables and weekly totals, stack, join and reshape tables;
  • show the answer with line, bar, scatter and histogram charts whose titles, axes and colors tell the truth;
  • make it repeatable and trustworthy: a notebook that tells a story, logic in a module, a script that rebuilds the report, and tests that prove its numbers and charts.

Here is one question that uses several of them at once, on the Greenway Food Truck's eight weeks of sales: how much did each location take in, week by week?

from pathlib import Path

import pandas as pd

weeks = [pd.read_csv(path).assign(week=path.stem[5:]) for path in sorted(Path("data/incoming").glob("week-*.csv"))]
sales = pd.concat(weeks, ignore_index=True)
sales["revenue"] = sales["quantity"] * sales["price"]
table = sales.pivot_table(index="week", columns="location", values="revenue", aggfunc="sum")
table.round(0)
location    City Park  Office Row  Stadium
week
2026-09-07      934.0      1783.0   1936.0
2026-09-14     1130.0      1873.0   1761.0
2026-09-21     1116.0      1503.0   1810.0
2026-09-28     1060.0      1928.0   1802.0
2026-10-05      968.0      1921.0   1622.0
2026-10-12      985.0      1852.0   1591.0
2026-10-19     1066.0      1883.0   1689.0
2026-10-26     1018.0      1706.0   1730.0

Files found with pathlib, labeled by assign and stacked, a new column from two others, and a pivot table: none of it would have made sense on your first day, and now you can read every line. The answer is worth a second look, too: every week, City Park takes in far less than Office Row or the Stadium.

Try it: a report on every week

Lessons and exercises hand you the problem. Here is one with no step-by-step instructions. Use the eight weeks from Exercise: Test the Food Truck Report and build a report across all of them:

  • Revenue week by week for each location, as a line chart whose title states what you found.
  • Each location's best and worst week, in a small table.
  • The best item of every week, and whether it ever changed.
  • One command that rebuilds the whole report as a web page when a ninth week arrives (python seed.py --next adds one).
  • Tests for every function that makes a number, on small weeks you worked out by hand.

There's no test suite for this one: when your own tests pass on small weeks you worked out by hand, it works.

An analysis of your own

The step after that is the one every analyst takes: choosing the question and the data yourself. Use one of the course's datasets, or your own data (a budget, a running log, a club's attendance, with nothing personal in it), and give it what a trustworthy analysis needs:

  • three to five questions, written down before you start;
  • the cleaning, in one cleaning function with checks, and a note of what it changed and why;
  • an answer to each question from a function in a module, with pytest tests for each one;
  • at least three charts that fit their questions, with titles that state the finding and labels with units;
  • a story notebook, run from top to bottom and shared as a web page, that says what the data can't tell us;
  • a README that says where the data came from and how to rebuild everything.

Start small: one question answered and tested, then the next. A finished analysis with a README is something you can show a teacher, a hiring manager or a friend.

Keep your data skills growing

  • Ask before you load. Write the question down first; it decides which columns matter and which cleaning is honest.
  • Count before and after every drop, filter and join. A number that changes when you didn't expect it is a question to answer.
  • Look, then prove. Explore in a notebook, then move what works into a function with a test.
  • Say what the data can't show. A small sample, a missing month or a survey only some people answered belongs in the report.

Courses that go further

When you know which direction pulls you, a course of its own takes you further. Each one stands alone and starts from the beginning of its subject:

  • Data Structures and Big-O, Demystified: why some code stays fast as the data grows and some crawls, with lists, stacks and queues, hash tables, trees and graphs, every sample shown in Python first.
  • Bash in Depth: the terminal from your first command to scripts that are real tools: organizing data files, searching and reporting on them with pipes, and automating a project, on a Mac, on Linux, and on Windows through WSL.
  • PowerShell in Depth: the same journey on Windows: organizing files, turning spreadsheets and web data into reports, and scripts that are small applications.

Check yourself

2 questions to try on the lesson page.

So what

You can turn a messy file into an answer people can trust and rebuild; an analysis of your own is where that turns into something you can show.

Key terms

Related