Can you advise if this is reasonable usage of

parameterize

and

inject

? This is getting me the result, but the column headers are using the name from the original

source("raw_fee_type")

Is this expected or am I using this incorrectly?

@parameterize(**{
    f"is_{exec_type}": {"raw_fee_type": source("raw_fee_type"), "search_exec_type": value(exec_type)}
    for exec_type in ["tas", "block", "efp", "efr", "eoo"]
})
def extract_execution_types(raw_fee_type: pd.Series, search_exec_type: str) -> pd.Series:
    return raw_fee_type.str.contains(search_exec_type, na=False)

@inject(
    all_execution_types=group(*[
        source(f"is_{exec_type}")
        for exec_type in ["tas", "block", "efp", "efr", "eoo"]
        
    ])
)
def execution_types(all_execution_types: List[pd.Series]) -> pd.DataFrame:
    return pd.concat(all_execution_types, axis="columns")

So this is right, but fairly heavy handed. Re: column-headers, I think this is about being explicit about labeling. It will use the node names to join as columns — you’ll need to set that correctly or rename them. That said, you’re getting to the point where you may want to think a little simpler, or maybe use a higher-level constrcut such as subdag. You’re doing something like this: 1. Have multiple different columns 2. Do a set of operations on them 3. Join them Those could be dataframe operations if you want, or you could run a subdag of operations on each column, then join later

Yeah the renaming fixed this approach. May think more about subdag for simplicity.

raw_fee_type.str.contains(search_exec_type, na=False).rename(f"is_{search_exec_type}")