Hands‑On NumPy Project: Synthetic Sales Dataset

Hands‑On NumPy Project

In this project you will create a synthetic sales dataset, clean it, and compute key metrics using NumPy.

Step 1: Generate synthetic data

import numpy as np
rng = np.random.default_rng(2024)

n = 1000
# Columns: [store_id, product_id, units_sold, price]
store_ids = rng.integers(1, 11, size=n)          # 10 stores
product_ids = rng.integers(100, 200, size=n)      # 100 products
units = rng.poisson(lam=20, size=n)               # realistic sales volume
price = rng.uniform(5, 100, size=n).round(2)

sales = np.column_stack((store_ids, product_ids, units, price))
print(sales[:5])

Step 2: Compute total revenue per store

revenue = sales[:,2] * sales[:,3]
# Aggregate revenue per store using np.bincount (store IDs start at 1)
rev_per_store = np.bincount(store_ids, weights=revenue)
print('Revenue per store (index = store_id):', rev_per_store[1:])

Step 3: Identify top‑selling product

units_per_product = np.bincount(product_ids, weights=units)
top_product = product_ids[np.argmax(units_per_product)]
print('Top selling product ID:', top_product)

Step 4: Visual sanity check (using Matplotlib later)

import matplotlib.pyplot as plt
plt.bar(range(1, len(rev_per_store)), rev_per_store[1:])
plt.xlabel('Store ID')
plt.ylabel('Revenue ($)')
plt.title('Revenue per Store')
plt.show()

What you learned

• Creating synthetic data with NumPy's random API.
• Vectorized arithmetic for revenue.
• Aggregating using np.bincount.
• Preparing data for downstream visualizations.