Python:Pandas 向量化运算替代循环,百万级数据提速实战

import pandas as pdimport numpy as np# 生成10万行超大测试数据集df = pd.DataFrame({    "goods_id": range(1, 100001),    "cost_price": np.random.uniform(30, 1200, 100000).round(2),    "sale_price": np.random.uniform(80, 3600, 100000).round(2),    "loss_rate": np.random.uniform(0, 0.15, 100000).round(3),    "stock_qty": np.random.randint(10, 500, 100000)})df.to_excel("vector_goods_data.xlsx", index=False)print("十万行库存商品测试数据生成完成")

import pandas as pdimport numpy as npimport timedf = pd.read_excel("vector_goods_data.xlsx")# 方案1：传统for循环（低效，仅作对比）start_loop = time.time()profit_loop = []for idx, row in df.iterrows():    real_sale = row["sale_price"] * (1 - row["loss_rate"])    single_profit = real_sale - row["cost_price"]    profit_loop.append(single_profit)df["loop_single_profit"] = profit_loopend_loop = time.time()# 方案2：向量化运算（高性能，生产推荐）start_vec = time.time()df["real_sale_unit"] = df["sale_price"] * (1 - df["loss_rate"])df["vec_single_profit"] = df["real_sale_unit"] - df["cost_price"]# 向量化多条件判定：区分盈利/亏损商品df["profit_tag"] = np.where(df["vec_single_profit"] > 0, "盈利", "亏损")df["total_profit"] = df["vec_single_profit"] * df["stock_qty"]end_vec = time.time()print(f"for循环执行耗时：{round(end_loop - start_loop, 2)}秒")print(f"向量化运算执行耗时：{round(end_vec - start_vec, 2)}秒")print("\n数据预览：")print(df[["goods_id","vec_single_profit","profit_tag","total_profit"]].head(8))