Topper and average for each subject from CSV rows — interview question

Each row is name,subject,marks. For every subject, in alphabetical order, print:

the topper: the highest mark; if two students tie, the alphabetically smaller name;
the topper's mark;
the average mark for the subject, with exactly 2 decimal places.

Constraints

1 ≤ n ≤ 10^4

Input and output

Input: First line: n. Then n lines `name,subject,marks` (marks are integers from 0 to 100; no spaces).
Output: One line per subject: `subject topper marks average`.

Examples

Input 1

6
Riya,maths,91
Dev,maths,91
Arjun,physics,78
Riya,physics,64
Dev,dsa,88
Tara,maths,70

Output

dsa Dev 88 88.00
maths Dev 91 84.00
physics Arjun 78 71.00

Dev and Riya tie on 91 in maths; Dev comes first alphabetically. The maths average is (91 + 91 + 70) / 3 = 84.00.

Input 2

2
Sana,os,0
Neha,os,1

Output

os Neha 1 0.50

A 0 is a valid mark. The average is 0.50.

Show hints

Keep one running record per subject: the best (marks, name), the total and the count.
To compare "highest mark, then smallest name" with one max, use the key (marks, reversed name order). Or compare tuples by hand.

Show the explanation

Make one pass and keep a small record per subject:

Python

best = {}                       # subject -> (marks, name)
total, count = defaultdict(int), defaultdict(int)
for row in rows:
    name, subject, m = row.split(",")
    m = int(m)
    total[subject] += m
    count[subject] += 1
    cur = best.get(subject)
    if cur is None or m > cur[0] or (m == cur[0] and name < cur[1]):
        best[subject] = (m, name)

Then for each subject in sorted(best), print the topper and f"{total[subject] / count[subject]:.2f}".

Discussion points

One pass, O(n) memory per subject at most. The rows do not need to be stored, so the same code works on a file streamed line by line: for line in open(path).
Tie-breaking. The rule "highest marks, then smallest name" mixes descending and ascending order. Write the comparison explicitly, or sort with key=lambda r: (-r.marks, r.name) and take the first row.
Real CSV files can contain quoted commas. Then use the csv module (csv.reader) rather than split(","). In data work, pandas.read_csv with groupby("subject").agg(...) does this in two lines. Interviewers often ask for both versions.
Formatting: :.2f rounds the floating-point value. Values exactly halfway between two outputs can depend on binary representation. For money, use decimal.Decimal.

Original UniCircle question. Last reviewed 22 Oct 2026.