def _compute_rater_factors()

in sourcecode/scoring/pflip_model.py [0:0]


  def _compute_rater_factors(self, prescoringRaterModelOutput: pd.DataFrame) -> pd.DataFrame:
    """Generate a DataFrame mapping raterParticipantIds to factors.

    Each rater is assigned their factor from either Core, Expansion or ExpansionPlus,
    prioritizing Core, then Expansion, then ExpansionPlus.

    Args:
      prescoringRaterModelOutput: pd.DataFrame used to determine rater factors.

    Returns:
      pd.DataFrame with raterParticipantId and RATER_FACTOR columns
    """
    # Obtain prescoring rater factors
    coreFactors = prescoringRaterModelOutput[
      prescoringRaterModelOutput[c.scorerNameKey] == Scorers.MFCoreScorer.name
    ][[c.raterParticipantIdKey, c.internalRaterFactor1Key]].rename(
      columns={c.internalRaterFactor1Key: c.coreRaterFactor1Key}
    )
    expansionFactors = prescoringRaterModelOutput[
      prescoringRaterModelOutput[c.scorerNameKey] == Scorers.MFExpansionScorer.name
    ][[c.raterParticipantIdKey, c.internalRaterFactor1Key]].rename(
      columns={c.internalRaterFactor1Key: c.expansionRaterFactor1Key}
    )
    expansionPlusFactors = prescoringRaterModelOutput[
      prescoringRaterModelOutput[c.scorerNameKey] == Scorers.MFExpansionPlusScorer.name
    ][[c.raterParticipantIdKey, c.internalRaterFactor1Key]].rename(
      columns={c.internalRaterFactor1Key: c.expansionPlusRaterFactor1Key}
    )
    # Combine and prioritize factors
    raterFactors = coreFactors.merge(expansionFactors, how="outer").merge(
      expansionPlusFactors, how="outer"
    )
    raterFactors[_RATER_FACTOR] = raterFactors[c.expansionPlusRaterFactor1Key]
    raterFactors.loc[
      ~raterFactors[c.expansionRaterFactor1Key].isna(), _RATER_FACTOR
    ] = raterFactors[c.expansionRaterFactor1Key]
    raterFactors.loc[~raterFactors[c.coreRaterFactor1Key].isna(), _RATER_FACTOR] = raterFactors[
      c.coreRaterFactor1Key
    ]
    return raterFactors[[c.raterParticipantIdKey, _RATER_FACTOR]]