DOI

The article discusses the use of textual data (company description)
to assign a fair market capitalization of an enterprise. The use of a simple document representation – a bag of words model, coupled with the use of gradient boosted decision trees (GBDT) allows generate precise business valuations. The test set rsquared for the NYSE and NASDAQ companies reaches 55.87%, and the average absolute percentage error of the forecast is only 9.58%. Thus, textual data has great potential for business valuation.
Translated title of the contributionBUSINESS VALUATION USING TEXT DATA
Original languageRussian
Title of host publicationСовременная экономика и право: опыт теоретического и эмпирического анализа
Subtitle of host publicationсборник статей III Международной научно-практической конференции
Place of PublicationПетрозаводск
PublisherНовая наука
Pages27-35
ISBN (Print)9785001746683
DOIs
StatePublished - 25 Aug 2022
EventСовременная экономика и право: опыт теоретического и эмпирического анализа: III Международная научно-практическая конференция - Петрозаводск, Russian Federation
Duration: 25 Aug 2022 → 25 Aug 2022

Conference

ConferenceСовременная экономика и право: опыт теоретического и эмпирического анализа
Country/TerritoryRussian Federation
CityПетрозаводск
Period25/08/22 → 25/08/22

    Research areas

  • business valuation, text data, nlp, bag of words, machine learning, GBDT, gradient boosting

ID: 103051066