Part 16 — Reshape & Merge

Fundamental Series — Part 16 of 20

Dua operasi penting: reshape (wide ↔︎ long) dan merge (gabung tabel).

Wide → Long

Pandas — `melt()`

import pandas as pd

df_wide = pd.DataFrame({
    "nama": ["Andi", "Budi"],
    "math": [80, 90],
    "sains": [75, 85],
    "bahasa": [88, 92]
})

df_long = df_wide.melt(
    id_vars="nama",
    var_name="mata_pelajaran",
    value_name="nilai"
)

Polars — `unpivot()`

import polars as pl

df_wide_pl = pl.DataFrame({
    "nama": ["Andi", "Budi"],
    "math": [80, 90],
    "sains": [75, 85],
    "bahasa": [88, 92]
})

df_long_pl = df_wide_pl.unpivot(
    index="nama",
    variable_name="mata_pelajaran",
    value_name="nilai"
)

Long → Wide

Pandas — `pivot()`

df_long.pivot(
    index="nama",
    columns="mata_pelajaran",
    values="nilai"
).reset_index()

Polars — `pivot()`

df_long_pl.pivot(
    on="mata_pelajaran",
    index="nama",
    values="nilai"
)

Merge / Join

siswa = pd.DataFrame({"id": [1,2,3,4], "nama": ["Andi","Budi","Citra","Dina"]})
nilai = pd.DataFrame({"id": [1,2,3,5], "skor": [80,90,75,85]})

Pandas

# Left join
siswa.merge(nilai, on="id", how="left")

# Inner join
siswa.merge(nilai, on="id", how="inner")

# Full / outer join
siswa.merge(nilai, on="id", how="outer")

# Kolom berbeda nama
siswa.merge(nilai, left_on="id", right_on="id")

Polars

siswa_pl = pl.DataFrame({"id": [1,2,3,4], "nama": ["Andi","Budi","Citra","Dina"]})
nilai_pl = pl.DataFrame({"id": [1,2,3,5], "skor": [80,90,75,85]})

siswa_pl.join(nilai_pl, on="id", how="left")
siswa_pl.join(nilai_pl, on="id", how="inner")
siswa_pl.join(nilai_pl, on="id", how="full")
siswa_pl.join(nilai_pl, on="id", how="anti")    # tidak match

Concat — Tumpuk Data

# Pandas
df1 = pd.DataFrame({"nama": ["A","B"], "x": [1,2]})
df2 = pd.DataFrame({"nama": ["C","D"], "x": [3,4]})
pd.concat([df1, df2], ignore_index=True)

# Polars
df1_pl = pl.DataFrame({"nama": ["A","B"], "x": [1,2]})
df2_pl = pl.DataFrame({"nama": ["C","D"], "x": [3,4]})
pl.concat([df1_pl, df2_pl])

Latihan

Latihan 16.1 — Reshape

gdp = pd.DataFrame({
    "negara": ["Indonesia", "Malaysia", "Thailand"],
    "2020": [1060, 337, 500],
    "2021": [1186, 373, 506],
    "2022": [1319, 407, 536]
})

# 1. Melt ke long (kolom: negara, tahun, gdp)
# 2. Pivot kembali ke wide
# Kerjakan dengan pandas DAN polars

Latihan 16.2 — Join

produk = pd.DataFrame({"id": [1,2,3], "nama": ["A","B","C"]})
penjualan = pd.DataFrame({"id": [1,1,2,4], "jumlah": [10,20,15,5]})

# 1. Left join
# 2. Hanya produk tanpa penjualan (anti join)
# 3. Inner join

Ringkasan

Operasi	Pandas	Polars
Wide → Long	`.melt()`	`.unpivot()`
Long → Wide	`.pivot()`	`.pivot()`
Join	`.merge()`	`.join()`
Concat	`pd.concat()`	`pl.concat()`

Sebelumnya: Part 15 — Group & Summarise Selanjutnya: Part 17 — Apply & Vectorisasi

--- title: "Part 16 — Reshape & Merge" description: "Mengubah bentuk data dan menggabungkan tabel di Python: melt, pivot, merge, dan concat dengan pandas dan polars." date: "2026-02-26" categories: [Fundamental, Reshape] series: "Fundamental Programming" order: 16 --- ## Fundamental Series — Part 16 of 20 Dua operasi penting: **reshape** (wide ↔ long) dan **merge** (gabung tabel). --- ## Wide → Long ### Pandas — `melt()` ```python import pandas as pd df_wide = pd.DataFrame({ "nama": ["Andi", "Budi"], "math": [80, 90], "sains": [75, 85], "bahasa": [88, 92] }) df_long = df_wide.melt( id_vars="nama", var_name="mata_pelajaran", value_name="nilai" ) ``` ### Polars — `unpivot()` ```python import polars as pl df_wide_pl = pl.DataFrame({ "nama": ["Andi", "Budi"], "math": [80, 90], "sains": [75, 85], "bahasa": [88, 92] }) df_long_pl = df_wide_pl.unpivot( index="nama", variable_name="mata_pelajaran", value_name="nilai" ) ``` --- ## Long → Wide ### Pandas — `pivot()` ```python df_long.pivot( index="nama", columns="mata_pelajaran", values="nilai" ).reset_index() ``` ### Polars — `pivot()` ```python df_long_pl.pivot( on="mata_pelajaran", index="nama", values="nilai" ) ``` --- ## Merge / Join ```python siswa = pd.DataFrame({"id": [1,2,3,4], "nama": ["Andi","Budi","Citra","Dina"]}) nilai = pd.DataFrame({"id": [1,2,3,5], "skor": [80,90,75,85]}) ``` ### Pandas ```python # Left join siswa.merge(nilai, on="id", how="left") # Inner join siswa.merge(nilai, on="id", how="inner") # Full / outer join siswa.merge(nilai, on="id", how="outer") # Kolom berbeda nama siswa.merge(nilai, left_on="id", right_on="id") ``` ### Polars ```python siswa_pl = pl.DataFrame({"id": [1,2,3,4], "nama": ["Andi","Budi","Citra","Dina"]}) nilai_pl = pl.DataFrame({"id": [1,2,3,5], "skor": [80,90,75,85]}) siswa_pl.join(nilai_pl, on="id", how="left") siswa_pl.join(nilai_pl, on="id", how="inner") siswa_pl.join(nilai_pl, on="id", how="full") siswa_pl.join(nilai_pl, on="id", how="anti") # tidak match ``` --- ## Concat — Tumpuk Data ```python # Pandas df1 = pd.DataFrame({"nama": ["A","B"], "x": [1,2]}) df2 = pd.DataFrame({"nama": ["C","D"], "x": [3,4]}) pd.concat([df1, df2], ignore_index=True) # Polars df1_pl = pl.DataFrame({"nama": ["A","B"], "x": [1,2]}) df2_pl = pl.DataFrame({"nama": ["C","D"], "x": [3,4]}) pl.concat([df1_pl, df2_pl]) ``` --- ## Latihan ::: {.callout-caution} ## Latihan 16.1 — Reshape ```python gdp = pd.DataFrame({ "negara": ["Indonesia", "Malaysia", "Thailand"], "2020": [1060, 337, 500], "2021": [1186, 373, 506], "2022": [1319, 407, 536] }) # 1. Melt ke long (kolom: negara, tahun, gdp) # 2. Pivot kembali ke wide # Kerjakan dengan pandas DAN polars ``` ::: ::: {.callout-caution} ## Latihan 16.2 — Join ```python produk = pd.DataFrame({"id": [1,2,3], "nama": ["A","B","C"]}) penjualan = pd.DataFrame({"id": [1,1,2,4], "jumlah": [10,20,15,5]}) # 1. Left join # 2. Hanya produk tanpa penjualan (anti join) # 3. Inner join ``` ::: --- ## Ringkasan | Operasi | Pandas | Polars | |---------|--------|--------| | Wide → Long | `.melt()` | `.unpivot()` | | Long → Wide | `.pivot()` | `.pivot()` | | Join | `.merge()` | `.join()` | | Concat | `pd.concat()` | `pl.concat()` | --- *Sebelumnya: [Part 15 — Group & Summarise](../2026_02_26_15_group_summarise/index.qmd)* *Selanjutnya: [Part 17 — Apply & Vectorisasi](../2026_02_26_17_apply/index.qmd)*