如何在Polars中基于列条件筛选计算累积和并生成EWM_COLUMN

我有一个Polars DataFrame，想要获取集群内球队的最新进球数来应用ewm_mean()，最终生成示例中的EWM_COLUMN列。

示例数据代码 #

import polars as pl

pl.Config(tbl_cols=9)

df = pl.read_csv(b"""
Season,Wk,Home,Away,HomeGoals,AwayGoals,Cluster_home,Cluster_away,Cluster_pair_key
2024,27.0,teamA,teamF,3,2,4,1,1_4
2024,27.0,teamB,teamG,1,3,2,2,2_2
2024,27.0,teamC,teamH,1,0,5,3,3_5
2024,27.0,teamD,teamI,0,1,3,1,1_3
2024,27.0,teamE,teamJ,3,0,3,4,3_4
""")

原始表格 #

┌────────┬──────┬───────┬───────┬───────────┬───────────┬──────────────┬──────────────┬──────────────────┐
│ Season ┆ Wk   ┆ Home  ┆ Away  ┆ HomeGoals ┆ AwayGoals ┆ Cluster_home ┆ Cluster_away ┆ Cluster_pair_key │
│ ---    ┆ ---  ┆ ---   ┆ ---   ┆ ---       ┆ ---       ┆ ---          ┆ ---          ┆ ---              │
│ i64    ┆ f64  ┆ str   ┆ str   ┆ i64       ┆ i64       ┆ i64          ┆ i64          ┆ str              │
╞════════╪══════╪═══════╪═══════╪═══════════╪═══════════╪══════════════╪══════════════╪══════════════════╡
│ 2024   ┆ 27.0 ┆ teamA ┆ teamF ┆ 3         ┆ 2         ┆ 4            ┆ 1            ┆ 1_4              │
│ 2024   ┆ 27.0 ┆ teamB ┆ teamG ┆ 1         ┆ 3         ┆ 2            ┆ 2            ┆ 2_2              │
│ 2024   ┆ 27.0 ┆ teamC ┆ teamH ┆ 1         ┆ 0         ┆ 5            ┆ 3            ┆ 3_5              │
│ 2024   ┆ 27.0 ┆ teamD ┆ teamI ┆ 0         ┆ 1         ┆ 3            ┆ 1            ┆ 1_3              │
│ 2024   ┆ 27.0 ┆ teamE ┆ teamJ ┆ 3         ┆ 0         ┆ 3            ┆ 4            ┆ 3_4              │
└────────┴──────┴───────┴───────┴───────────┴───────────┴──────────────┴──────────────┴──────────────────┘

需求说明 #

以teamE为例，计算其EWM_COLUMN时需要纳入以下进球数：

• teamE的HomeGoals
• teamD的HomeGoals
• teamC的AwayGoals

我已经创建了Cluster_pair_key_Organized列来辅助实现需求，最终希望生成如下表格中的EWM_COLUMN：

目标表格 #

内容的提问来源于stack exchange，提问作者Pedro_Siqueira