Part 17 — Apply & Vectorisasi

Operasi vectorized dan apply di Python: numpy vectorization, apply, map, list comprehension, dan kapan pakai loop.
Fundamental
Apply
Diterbitkan

26 Februari 2026

Fundamental Series — Part 17 of 20

Python (dengan NumPy/Polars) mendukung vectorized operations — operasi pada seluruh array sekaligus tanpa loop. Part ini membahas kapan vectorized vs apply vs loop.

Vectorized dengan NumPy

import numpy as np

x = np.array([1, 2, 3, 4, 5])

# Sudah vectorized — tanpa loop
x * 2          # array([2, 4, 6, 8, 10])
x ** 2         # array([1, 4, 9, 16, 25])
np.sqrt(x)     # array([1., 1.41, 1.73, 2., 2.24])
x > 3          # array([False, False, False, True, True])
np.where(x > 3, "besar", "kecil")

apply() di Pandas

import pandas as pd

df = pd.DataFrame({
    "nama": ["A", "B", "C"],
    "x": [10, 20, 30],
    "y": [100, 200, 300]
})

# Apply ke kolom (axis=0)
df[["x", "y"]].apply(np.mean)

# Apply ke baris (axis=1)
df[["x", "y"]].apply(sum, axis=1)

# Apply fungsi custom
df["x"].apply(lambda val: val ** 2)
Pentingapply() Pandas Lambat

apply() iterasi baris per baris — jauh lebih lambat dari vectorized. Gunakan hanya jika tidak ada alternatif vectorized.

map() dan map_elements() di Polars

import polars as pl

df_pl = pl.DataFrame({
    "nama": ["A", "B", "C"],
    "x": [10, 20, 30]
})

# Vectorized — CEPAT
df_pl.with_columns(
    (pl.col("x") ** 2).alias("x_squared")
)

# map_elements — LAMBAT (hanya jika terpaksa)
df_pl.with_columns(
    pl.col("x").map_elements(lambda v: v ** 2, return_dtype=pl.Int64).alias("x_squared")
)

List Comprehension — Pythonic Apply

# Alternatif apply untuk list biasa
data = [1, 2, 3, 4, 5]

[x ** 2 for x in data]                    # [1, 4, 9, 16, 25]
[x for x in data if x > 3]                # [4, 5]
[f"item_{x}" for x in data]               # ['item_1', ...]

# Dict comprehension
{k: v ** 2 for k, v in zip(["a","b","c"], [1,2,3])}

map() dan filter() Bawaan Python

# map
list(map(lambda x: x ** 2, [1, 2, 3, 4, 5]))
# [1, 4, 9, 16, 25]

# filter
list(filter(lambda x: x > 3, [1, 2, 3, 4, 5]))
# [4, 5]

Perbandingan Kecepatan

import numpy as np

x = np.random.randn(1_000_000)

# LAMBAT — Python loop
result = [xi ** 2 for xi in x]

# CEPAT — NumPy vectorized
result = x ** 2
TipAturan Praktis
  1. NumPy/Polars vectorized — tercepat
  2. List comprehension — cepat untuk list Python biasa
  3. apply() — lambat, gunakan terakhir

Latihan

BahayaLatihan 17.1 
# 1. Buat fungsi standarisasi(x) = (x - mean) / std
# 2. Apply ke dictionary {"a": [1..10], "b": [20..30]}
# 3. Di polars DataFrame, standarisasi semua kolom numerik (vectorized)

Ringkasan

Metode Speed Keterangan
NumPy vectorized ⚡⚡⚡ Tercepat
Polars expressions ⚡⚡⚡ Tercepat
List comprehension ⚡⚡ Bagus untuk list Python
map() built-in ⚡⚡ Functional style
pd.apply() ⚡ Lambat — hindari jika bisa

Sebelumnya: Part 16 — Reshape & Merge Selanjutnya: Part 18 — Debugging & Error Handling