Tugas top 5 brand, Desember 2019

from pathlib import Path
import matplotlib.pyplot as plt
import pandas as pd

pandas baca dan olah tabel, matplotlib.pyplot gambar grafik, pathlib cari CSV relatif ke posisi skrip. Matplotlib baru dipakai dari tugas 2.

csv_path = Path(__file__).resolve().parents[2] / "retail_raw_reduced.csv"
dataset = pd.read_csv(csv_path)

Path(__file__) = lokasi skrip. parents[2] naik dua folder ke sains-data, tempat retail_raw_reduced.csv. pd.read_csv() muat isinya ke dataset.

dataset['order_date'] = pd.to_datetime(dataset['order_date'])
dataset['order_month'] = dataset['order_date'].dt.strftime('%Y-%m')

order_date di-parse ke datetime, lalu order_month disimpan sebagai YYYY-MM (misalnya 2019-12). Filter Desember cukup order_month == '2019-12'.

top_brands = (dataset[dataset['order_month'] == '2019-12'].groupby('brand')['quantity']
              .sum()
              .reset_index()
              .sort_values(by='quantity', ascending=False)
              .head(5))

Jumlahkan quantity Desember 2019 per brand, urutkan menurun, head(5) ambil lima teratas.

dataset_top5brand_dec = dataset[(dataset['order_month'] == '2019-12') &
                                (dataset['brand'].isin(top_brands['brand'].to_list()))]
print(top_brands)

dataset_top5brand_dec = transaksi Desember dari brand-brand itu saja. print(top_brands) cetak brand dan total quantity.

top_brands = (dataset[dataset['order_month'] == '2019-12'].groupby('brand')['quantity']
              .sum()
              .reset_index()
              .sort_values(by='quantity', ascending=False)
              .head(5))
dataset_top5brand_dec = dataset[(dataset['order_month'] == '2019-12') &
                                (dataset['brand'].isin(top_brands['brand'].to_list()))]

Filter brand: copy dari tugas 1.

dataset_top5brand_dec.groupby(['order_date','brand'])['quantity'].sum().unstack().plot(marker='.', cmap='plasma')

groupby order_date dan brand, jumlahkan quantity. unstack() ubah tiap brand jadi kolom, jadi plot keluar lima garis. marker="." titik harian, cmap="plasma" beda warna per brand.

plt.title('Daily Sold Quantity Dec 2019 - Breakdown by Brands', loc='center', pad=30, fontsize=15, color='blue')
plt.xlabel('Order Date', fontsize=12)
plt.ylabel('Quantity', fontsize=12)
plt.grid(color='darkgray', linestyle=':', linewidth=0.5)
plt.ylim(ymin=0)
plt.legend(loc='upper center', bbox_to_anchor=(1.1, 1), shadow=True, ncol=1)

Judul dan label ikut slide. Grid putus-putus bantu baca skala. y mulai dari 0. Legend di kanan atas (bbox_to_anchor=(1.1, 1)) biar tidak nutup garis.

plt.annotate('Terjadi lonjakan', xy=(7, 310), xytext=(8, 300),
             weight='bold', color='red',
             arrowprops=dict(arrowstyle='->',
                             connectionstyle="arc3",
                             color='red'))

Panah merah ke (7, 310), sesuai lonjakan di slide (sekitar tanggal 8, quantity ~310). Teks "Terjadi lonjakan" di (8, 300).

plt.gcf().set_size_inches(10, 5)
plt.tight_layout()
plt.show()

Figure 10 × 5 inci. tight_layout rapikan ruang judul, label, legend.

top_brands = (dataset[dataset['order_month'] == '2019-12'].groupby('brand')['quantity']
              .sum()
              .reset_index()
              .sort_values(by='quantity', ascending=False)
              .head(5))
dataset_top5brand_dec = dataset[(dataset['order_month'] == '2019-12') &
                                (dataset['brand'].isin(top_brands['brand'].to_list()))]

Blok filter brand sama seperti tugas 1.

plt.clf()
dataset_top5brand_dec.groupby("brand")["product_id"].nunique().sort_values(ascending=False).plot(
    kind="bar",
    color='green')

plt.clf() kosongkan figure. groupby brand, nunique() hitung product_id unik, urut menurun, plot bar hijau.

plt.title("Jumlah Unique Product Terjual - Top 5 Brands (Desember 2019)", loc='center', fontsize=15, pad=30, color='blue')
plt.xlabel("Brand", fontsize=15)
plt.ylabel("Jumlah Unique Product", fontsize=15)
plt.ylim(ymin=0)
plt.xticks(rotation=0)
plt.show()

Judul pakai teks slide. y mulai 0, label brand x tidak diputar (rotation=0).

top_brands = (dataset[dataset['order_month'] == '2019-12'].groupby('brand')['quantity']
              .sum()
              .reset_index()
              .sort_values(by='quantity', ascending=False)
              .head(5))
dataset_top5brand_dec = dataset[(dataset['order_month'] == '2019-12') &
                                (dataset['brand'].isin(top_brands['brand'].to_list()))]

Filter brand ulang, sama seperti tugas 1.

dataset_top5brand_dec_per_product = dataset_top5brand_dec.groupby(['brand','product_id'])['quantity'].sum().reset_index()

groupby brand dan product_id, jumlahkan quantity. Satu baris = total quantity satu produk di satu brand selama Desember.

dataset_top5brand_dec_per_product['quantity_group'] = \
dataset_top5brand_dec_per_product['quantity'].apply(lambda x: '>= 100' if x >= 100 else '< 100')
dataset_top5brand_dec_per_product.sort_values('quantity', ascending=False, inplace=True)

quantity_group = '>= 100' atau '< 100' dari total quantity produk. sort_values urutkan dari quantity terbesar.

s_sort = dataset_top5brand_dec_per_product.groupby(['brand'])['product_id'].nunique().sort_values(ascending=False)

s_sort = jumlah product_id unik per brand, urut menurun. Urutan ini dipakai stacked bar, sama seperti tugas 3 (bukan alfabet).

dataset_top5brand_dec_per_product.groupby(['brand','quantity_group'])['product_id'].nunique().unstack().reindex(index=s_sort.index, level='brand').plot(kind='bar', stacked=True)

groupby brand dan quantity_group, nunique() hitung produk per kelompok. unstack() pisah >= 100 dan < 100 jadi kolom, reindex ikut s_sort, plot stacked bar.

plt.title('Number of Sold Products per Brand, December 2019', loc='center', pad=30, fontsize=15, color='blue')
plt.xlabel('Brand', fontsize=15)
plt.ylabel('Number of Products', fontsize=15)
plt.ylim(ymin=0)
plt.xticks(rotation=0)
plt.show()

Judul dan label ikut slide. y mulai 0, label brand x tidak diputar.