7.00.02 - Output - Aster Analytics

Teradata Aster® Analytics Foundation User GuideUpdate 2

Product
Aster Analytics
Release Number
7.00.02
Release Date
September 2017
Content Type
Programming Reference
User Guide
Publication ID
B700-1022-700K
Language
English (United States)
EvaluateSentimentExtractor Output Table Schema
Column Name Data Type Description
evaluation_result VARCHAR Reports these values:
  • Positive record (total relevant, relevant, total retrieved)
  • Recall and precision
  • Negative record (total relevant, relevant, total retrieved)
  • Recall and precision
  • Positive and negative record (total relevant, relevant, total retrieved)
  • Recall and precision

Given these definitions:

  • POS_EXPECT = count of rows in which observed sentiment is POS
  • POS_RETURN = count of rows in which actual sentiment is POS
  • POS_TRUE = count of rows in which both observed and actual sentiment are POS
  • NEG_EXPECT = count of rows in which observed sentiment is NEG
  • NEG_RETURN = count of rows in which actual sentiment is NEG
  • NEG_TRUE = count of rows in which both observed and actual sentiment are NEG
  • NEU_EXPECT = count of rows in which observed sentiment is NEU
  • NEU_RETURN = count of rows in which actual sentiment is NEU
  • NEU_TRUE = count of rows in which both observed and actual sentiment are NEU

Precision and recall are calculated as follows:

  • Positive sentiment:
    • Precision = POS_TRUE / POS_RETURN
    • Recall = POS_TRUE / POS_EXPECT
  • Negative sentiment:
    • Precision = NEG_TRUE / NEG_RETURN
    • Recall = NEG_TRUE / NEG_EXPECT
  • Neutral sentiment:
    • Precision = NEU_TRUE / NEU_RETURN
    • Recall = NEU_TRUE / NEU_EXPECT
  • All sentiment:
    • Precision = (POS_TRUE + NEG_TRUE + NEU_TRUE) / (POS_RETURN + NEG_RETURN + NEU_RETURN)
    • Recall = (POS_TRUE + NEG_TRUE + NEU_TRUE) / (POS_EXPECT + NEG_EXPECT + NEU_EXPECT)