NumPy: Percentiles with interpolation

np.percentile uses linear interpolation by default. Compute the same values.

What to build

A program that prints percentiles the way np.percentile does by default.

Requirements

  • For each requested q, in order, print pQ = V, with Q as given and V to two decimals.
  • Use the sorted data and position (n − 1) × q ÷ 100, interpolating linearly between the two neighbouring values.

Input

Line 1: the data, numbers separated by spaces. Line 2: the percentiles to compute, between 0 and 100.

Output

One line per percentile.

Constraints

1 to 1000 data values; 1 to 20 percentiles.

Example

Input:

1 2 3 4
25 50 75

Output:

p25 = 1.75
p50 = 2.50
p75 = 3.25

Why it works

Linear interpolation treats the sorted values as points spaced evenly from position 0 to n − 1 and joins them with straight lines, so every percentile has one well-defined value.

Examples

Example 1 · Input
1 2 3 4
25 50 75
Expected output
p25 = 1.75
p50 = 2.50
p75 = 3.25
Example 2 · Input
10
0 100
Expected output
p0 = 10.00
p100 = 10.00

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

Hints

Hint 1

Sort the data, then compute pos = (n − 1) × q ÷ 100.

Hint 2

Take the value at int(pos) and move the fraction pos − int(pos) of the way to the next value.

Solution

Show one correct solution (try it yourself first)
data = sorted(float(x) for x in input().split())
for q in input().split():
    pos = (len(data) - 1) * float(q) / 100
    low = int(pos)
    high = min(low + 1, len(data) - 1)
    value = data[low] + (pos - low) * (data[high] - data[low])
    print(f"p{q} = {value:.2f}")

Lesson: Statistics with NumPy →