Online Python classes in Göttingen

import csv

with open("goettingen_daily.csv", encoding="utf-8") as f:
    days = [row for row in csv.DictReader(f) if "2010" <= row["date"][:4] <= "2019"]

summer = 0
for day in days:
    if day["high"] != "-999" and float(day["high"]) >= 25:
        summer += 1

frost = 0
for day in days:
    if day["low"] != "-999" and float(day["low"]) < 0:
        frost += 1

wet = 0
for day in days:
    if day["low"] != "-999" and float(day["low"]) >= 1.0:
        wet += 1

print("2010s days:", len(days), " summer:", summer, " frost:", frost, " wet:", wet)

Does 2,647 wet days in ten years sound right? It is about 72% of all days. Many people would accept it without a second look. But look at the third block: it still reads the column "low", copied from the frost block, and only the rule was changed. It has counted nights that stayed above 1 degree, not rainy days. No error message, just a believable wrong number.

The fix is not to be more careful with copying. It is to stop copying. The Python tutorial puts it plainly: "The keyword def introduces a function definition." Version 2 writes the counting once, as a function that takes the days, the column and the rule, and calls it three times for every decade. The rule is passed as a small anonymous function; in the tutorial's words, "Small anonymous functions can be created with the lambda keyword."

def count(days, column, rule):
    total = 0
    for day in days:
        if day[column] != "-999" and rule(float(day[column])):
            total += 1
    return total

with open("goettingen_daily.csv", encoding="utf-8") as f:
    everything = list(csv.DictReader(f))

for start in range(1950, 2020, 10):
    decade = [d for d in everything if start <= int(d["date"][:4]) < start + 10]
    print(start, len(decade),
          count(decade, "high", lambda t: t >= 25),
          count(decade, "low", lambda t: t < 0),
          count(decade, "rain_mm", lambda r: r >= 1.0))

Counts are ours, from the DWD Climate Data Center daily file for station 01691 Göttingen (CC BY 4.0), fetched on 3 November 2026. The station metadata lists new positions on 15 December 1967, 12 August 1978, 10 November 1993 and 24 August 2007, so decades are not measured at exactly the same spot.

The real 2010s figure is 1,072 wet days, not 2,647. And with the function in place, the whole seventy-year table costs one loop. Any later fix, such as changing wet days to 5 mm or more, is made in one line and applies to every decade at once. That is why functions matter more than any other idea in the first months of Python.

1. Pupils, about 10 to 13. Before running Version 1, read the third block aloud and find the slip. Then run it and see whether you were right.
2. Teenagers. Add a fourth question, days of 30 degrees or more, with one line using count. Which decade had the most?
3. Adults and students. Rewrite count so it returns a share of days instead of a number, and explain why that is fairer for decades with 3,652 and 3,653 days.