Quick reference¶

Things to look up, in one place. Each one is taught properly somewhere in the weeks. This page is only the lookup.

When something breaks¶

Read the last line of the error first. It says what went wrong. The line number above it says where.

Error	What it usually means	Explained in
`NameError`	A name that was never defined, often a typo	Week 1
`TypeError`	Two types that do not fit. Usually text where a number was needed	Week 1
`IndexError`	You asked a list for a position it does not have	Week 1
`FileNotFoundError`	You are in the wrong folder. Check with `ls` or `dir`	Week 1
`IndentationError`	Lines do not line up. The program never started	Week 1
`ZeroDivisionError`	Something divided by zero, often an empty input	Week 1

The terminal¶

Command	What it does
`cd foldername`	Move into a folder
`cd ..`	Move up one folder
`ls` / `dir`	List this folder (macOS, Linux and Codespaces / Windows)
`python script.py`	Run a script
Ctrl-C	Stop a program that is not stopping by itself

Activate your environment in every new terminal. Without it, python is the wrong Python.

macOS, Linux, CodespacesWindows

source .venv/bin/activate

.venv\Scripts\activate

Your prompt shows (.venv) when it worked.

Python you will keep looking up¶

int("1703")            # text to whole number
float("3.1")           # text to decimal
f"Total: {total}"      # put a value inside text
f"{value:.1f}"         # one decimal place
f"{value:>8}"          # pad to 8 characters, right-aligned
f"{hour:02d}"          # pad a number to two digits: 8 becomes 08

counts[0]              # first item — positions start at zero
counts[-1]             # last item
counts[1:3]            # positions 1 and 2. The second number is excluded
[0] * 24               # a list of twenty-four zeros
len(counts)            # how many items

for value in counts:            # over items
for i in range(len(counts)):    # over positions
range(0, 24, 2)                 # 0, 2, 4 … 22

Reading pandas¶

You will read more of this than you write. Six lines cover most of it.

You see	Read it as
`len(df)`	how many rows
`df["speed"]`	one named column
`df[df["stop_id"] == "S001"]`	keep only the rows where this is true
`df["speed"].mean()`	one number summarising a column
`df.groupby("stop_id")["speed"].mean()`	one number per group
`left.merge(right, on="trip_id")`	join two tables on a shared column

One habit

Print len(df) before and after every filter and every join.

A filter that matches nothing gives an empty table. A merge on a repeated key multiplies rows. Neither prints a warning. Both give you a tidy table and a wrong answer.

Git, for keeping your progress¶

git status                       # what has changed
git add .                        # stage everything
git commit -m "chapter 2 works"  # save a snapshot
git log --oneline                # every snapshot so far
git restore atlas.py             # undo a file back to the last snapshot

Save a version every time something works. Then you can try something ambitious, because going back costs one line.

Project lookups you must not guess¶

You need	Where it comes from
ATCO area code	`project/data/external/atco_area_codes.csv`, all 150
PCT region name	The list in Data discovery. Historic counties, so Bristol is `avon`
Local-authority code	The deprivation file itself, beside the district name
ONS boundary service address	Copy it from Data discovery. Do not retype it

Code	Means
STATS19 `casualty_type` `0` / `1`	pedestrian / cyclist
STATS19 severity `1` / `2` / `3`	fatal / serious / slight
IMD decile `1`	the most deprived tenth of England
`-1` in a sensor feed	usually "no observation", not a measurement

The four checks¶

Applied to any code you did not write yourself.

Does it run?
Does it give the right answer on a case you already know? Work five rows out by hand. Compare.
What does it do with the awkward cases? Missing values, duplicates, zeros, empty inputs, the full file rather than the sample.
Can you explain every line?

The full version is the verification checklist. Print it.