如何用Pandas高效实现基于前置值的列赋值操作？

针对你需要的转换逻辑，核心是利用Pandas的向量化映射和前向填充能力替代循环，以下是关键方向：

• 第一步：处理明确映射关系
  用Series.replace()或np.where()直接将Event列中的Arrival、Departure转换为对应的At Sea/At Port，同时将其他值标记为缺失值（如NaN）。

• 第二步：追踪最近前置触发事件
  单独提取Event列中的Arrival/Departure，其余值设为NaN，再用Series.ffill()（前向填充）为每一行匹配到最近的前置触发事件，快速完成参考事件的全局关联。

• 第三步：处理非触发事件的映射
  基于前向填充得到的参考事件，用np.where()或Series.map()完成反转映射：参考事件为Arrival则映射为At Port，参考事件为Departure则映射为At Sea。

• 第四步：合并结果
  将明确映射的结果与非触发事件的映射结果合并，最终生成目标列。

全程采用Pandas向量级操作，相比循环效率提升明显，尤其适配大数据量场景。

附你的循环实现代码：

import knime.scripting.io as knio
import pandas as pd

# Create a pandas data frame with the contents of the KNIME node input table
inputTable = knio.input_tables[0].to_pandas()

# Resetting index for the main table to integer from 0 to n
inputTable.reset_index(drop=True, inplace=True)

# Create a series out of the colummn Event
eventSeries = inputTable.loc[:, "Event"]

# Create a new series with "none" values to use for writing the Sea/Port info into
length = eventSeries.size
seaPortValueSeries = pd.Series("none", index=range(length))
seaPortValueSeries = seaPortValueSeries.astype(str)

# Iterate over the elements of the series, writing a value into a second
# series that contains only "At Sea" and "At Port" as entries.
# For the first iteration, it makes no sense to check predecessors
# or retrieve predecessor values, so a shortened version is used for that iteration
for i in range(length):

    if i == 0:
        
        if eventSeries[i] == "Arrival":
            seaPortValueSeries[i] = "At_Sea"
        if eventSeries[i] == "Departure":
            seaPortValueSeries[i] = "In_Port"
    else:

        if eventSeries[i] == "Arrival":
            seaPortValueSeries[i] = "At_Sea"
        elif eventSeries[i] == "Departure":
            seaPortValueSeries[i] = "In_Port"
        elif eventSeries[i-1] == "Arrival":
            seaPortValueSeries[i] = "In_Port"
        elif eventSeries[i-1] == "Departure":
            seaPortValueSeries[i] = "At_Sea"
        else:
            seaPortValueSeries[i] = seaPortValueSeries[i-1]

# Append the second series as an additional column into the data frame
inputTable["Sea_Port"] = seaPortValueSeries

knio.output_tables[0] = knio.Table.from_pandas(inputTable)

内容的提问来源于stack exchange，提问作者schuppius