pandas: Work out each column's type

When pandas reads a file it gives each column one type. Model the rules it uses for the common cases.

What to build

A program that infers the dtype of each column of a comma-separated table.

Requirements

  • Empty cells are missing values. For the non-missing cells of a column: bool if all are True or False; int64 if all are whole numbers; float64 if all are numbers (some with a decimal point); otherwise object.
  • An int64 column with at least one missing value becomes float64; a column with no values at all is float64. A bool column with a missing value becomes object.
  • Print NAME: DTYPE (M missing) for each column in header order.

Input

Line 1: the header. Line 2: n. Then n rows, values separated by commas.

Output

One line per column.

Constraints

1 ≤ columns ≤ 10; 1 ≤ n ≤ 100; cells contain no commas.

Example

Input:

id,price,name,paid
3
1,9.5,Pen,True
2,,Book,False
3,4,,True

Output:

id: int64 (0 missing)
price: float64 (1 missing)
name: object (1 missing)
paid: bool (0 missing)

Why it works

A column holds one type, and NaN, pandas' missing value, is a float, which is why a single gap turns a whole integer column into floats, and why a missing value in a boolean column falls back to object.

Examples

Example 1 · Input
id,price,name,paid
3
1,9.5,Pen,True
2,,Book,False
3,4,,True
Expected output
id: int64 (0 missing)
price: float64 (1 missing)
name: object (1 missing)
paid: bool (0 missing)
Example 2 · Input
age,score
2
30,
,
Expected output
age: float64 (1 missing)
score: float64 (2 missing)

Prompts inside input() are ignored when checking, so input("n: ") is fine.

Hints

Hint 1

Check every non-empty cell: try int(), then float(), and test for True/False.

Hint 2

An int column with an empty cell becomes float64, because NaN is a float; an all-empty column is float64 too.

Solution

Show one correct solution (try it yourself first)
def is_int(text):
    try:
        int(text)
        return True
    except ValueError:
        return False


def is_float(text):
    try:
        float(text)
        return True
    except ValueError:
        return False


header = input().split(",")
n = int(input())
rows = [input().split(",") for _ in range(n)]
for i, name in enumerate(header):
    cells = [row[i] for row in rows]
    present = [c for c in cells if c != ""]
    missing = len(cells) - len(present)
    if not present:
        dtype = "float64"
    elif all(c in ("True", "False") for c in present):
        dtype = "object" if missing else "bool"
    elif all(is_int(c) for c in present):
        dtype = "float64" if missing else "int64"
    elif all(is_float(c) for c in present):
        dtype = "float64"
    else:
        dtype = "object"
    print(f"{name}: {dtype} ({missing} missing)")

Lesson: Why pandas, and how this course runs it →