Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Predicting taxi passenger demand using artificial neural networks
KTH, School of Computer Science and Communication (CSC).
2017 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesisAlternative title
Uppskattning av taxibeställningar med artificiella neurala nätverk (Swedish)
Abstract [en]

In this report a machine learning method using artificial neural networks to estimate taxi demand in different geographical zones in the city of Stockholm is proposed. An attempt to determine the most important input features that affect taxi ridership is performed and a network architecture is conceived and trained using taxi ridership data from a major taxi company operating in the city. The results show that except for the two basic input parameters, the hour of the day and the zone, the day of the week is clearly the most important factor. Also days after payment and month of the year seems to be mildly relevant factors while rain and temperature hardly affect the results at all. The final network model conceived was capable of estimating taxi demand in Stockholm with an average error of 2.73 rides and a success rate of 46 % of the rides using a boundary of 30 % or 1 ride.

Abstract [sv]

I den här rapporten utvecklas och utvärderas en metod för att uppskatta taxibehov i olika geografiska zoner i Stockholm med hjälp av artificiella neurala nätverk. Rapporten fokuserar på att ta reda på de mest relevanta parametrarna som påverkar taxiåkande. Dessa används som indata till ett neuralt nätverk som tränas med historisk data från ett av Stockholms största taxibolag. Resultaten visar hur vissa parametrar såsom timme på dygnet samt veckodag tydligt påverkar mängden taxiresor i olika områden medan andra parametrar såsom temperatur och nederbörd knappt påverkar taxiresande alls. Det slutliga valet av modell och inputparametrar lyckas förutspå korrekt antal körningar med ett snittfel på 2.73 körningar eller i 46 % av fallen i testdatan när man räknar en korrekt uppskattning att ligga som mest 30 % eller en körning ifrån det korrekta värdet.

Place, publisher, year, edition, pages
2017. , p. 51
Keywords [en]
taxi, neural, network
National Category
Computer Sciences
Identifiers
URN: urn:nbn:se:kth:diva-203368OAI: oai:DiVA.org:kth-203368DiVA, id: diva2:1082065
Subject / course
Computer Science
Educational program
Master of Science in Engineering - Computer Science and Technology
Supervisors
Examiners
Available from: 2017-03-16 Created: 2017-03-15 Last updated: 2018-01-13Bibliographically approved

Open Access in DiVA

fulltext(3175 kB)482 downloads
File information
File name FULLTEXT01.pdfFile size 3175 kBChecksum SHA-512
08efbbe808ad7b25ea18de9d5458c03cbb28765b7862e8691862a630ab659a0828470bb495ffb3cf9e5fbdbd17b07091aff73f51cf32308155dcc4d414de92e5
Type fulltextMimetype application/pdf

Search in DiVA

By author/editor
Zander, Gustav
By organisation
School of Computer Science and Communication (CSC)
Computer Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 482 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 530 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf