search query: @keyword koneoppiminen / total: 63
reference: 30 / 63
« previous | next »
Author:Airaksinen, Sami
Title:HIP: Model Combination Algorithm for Location Prediction
HIP: Kombinaatiomallinnus algoritmi paikkaennustukseen
Publication type:Master's thesis
Publication year:2014
Pages:69      Language:   eng
Department/School:Perustieteiden korkeakoulu
Main subject:Informaatiotekniikka   (T3006)
Supervisor:Oja, Erkki
Instructor:Aksela, Matti
Electronic version URL: http://urn.fi/URN:NBN:fi:aalto-201410052745
OEVS:
Electronic archive copy is available via Aalto Thesis Database.
Instructions

Reading digital theses in the closed network of the Aalto University Harald Herlin Learning Centre

In the closed network of Learning Centre you can read digital and digitized theses not available in the open network.

The Learning Centre contact details and opening hours: https://learningcentre.aalto.fi/en/harald-herlin-learning-centre/

You can read theses on the Learning Centre customer computers, which are available on all floors.

Logging on to the customer computers

  • Aalto University staff members log on to the customer computer using the Aalto username and password.
  • Other customers log on using a shared username and password.

Opening a thesis

  • On the desktop of the customer computers, you will find an icon titled:

    Aalto Thesis Database

  • Click on the icon to search for and open the thesis you are looking for from Aaltodoc database. You can find the thesis file by clicking the link on the OEV or OEVS field.

Reading the thesis

  • You can either print the thesis or read it on the customer computer screen.
  • You cannot save the thesis file on a flash drive or email it.
  • You cannot copy text or images from the file.
  • You cannot edit the file.

Printing the thesis

  • You can print the thesis for your personal study or research use.
  • Aalto University students and staff members may print black-and-white prints on the PrintingPoint devices when using the computer with personal Aalto username and password. Color printing is possible using the printer u90203-psc3, which is located near the customer service. Color printing is subject to a charge to Aalto University students and staff members.
  • Other customers can use the printer u90203-psc3. All printing is subject to a charge to non-University members.
Location:P1 Ark Aalto  1737   | Archive
Keywords:predictive analysis
machine learning
mixture
geolocation
prediktiivinen mallintaminen
koneoppiminen
mikstuurimallit
paikkatieto
Abstract (eng):In this thesis a working proof of concept of an algorithm that utilizes multiple regression algorithms and their temporal prediction performance has been developed and analyzed.
This framework was named history predictor (HIP).
Analysis of the algorithm was based on a temporal-spatial data set over 87 persons mobile location behavior.
The overall subject is large to cover and therefore this study concentrated on developing an algorithm with simple 'defaults' on cost of potentially better performance.

The framework consisted of three distinct phases.
First, data was processed to 67 statistical features by using a sliding window technique where each feature was generated over users' location history.
Curve fitting and statistical properties were used to extract features from the source data.
In the second phase pre-selected regression models were trained against the generated features.
Third phase, multiple different selection layer implementations were tried and their performance was analyzed against each other and the base regression models.
Part of the analysis was focused on the best selection layer and its performance with in depth analysis of its internals.

It was found that HIP increased performance over a single predicting model.
When compared to the best performing model, HIP gained relative improvement of 32 % on the hitrate performance compared against the best performing base model.
This was verified with Wilcoxon signed rank test to be statistically significant.
Therefore one can conclude that HIP should be considered a better approach for this data set and topic than using any of the trained regression models independently.
Abstract (fin):Tämän opinnäytetyön tavoitteena oli luoda toimiva konsepti koneoppimisalgoritmista, joka pohjautuu joukkoon regressiomalleja joiden ennustuksen tarkkuutta lähihistoriassa hyödynnettiin ennustuksen parantamiseen.
Analyysi tehtiin paikannustiedolla, jossa oli 87 eri käyttäjän puhelimen liikehistoria pitkältä aikaväliltä.

Kehitetty kehys sisälsi 3 erillistä vaihetta.
Ensimmäisessä vaiheessa tietojoukolle tehtiin 67 piirteenirrotusta siirtyvällä aikaikkuna tekniikalla.
Piirteidenirrotuksessa käytettiin käyräsovitusta ja tiedon statistisien ominaisuuksien määrittelyä.
Toisessa vaiheessa esivalitut regressiomallit koulutettiin piirteitä vasten.
Kolmannessa vaiheessa useita eri kombinaatiostrategioita analysoitiin ja niitä verrattiin yksittäisiin koulutettuihin regressiomalleihin.
Myös syväluotaava analyysi yhdistelmästrategioitten toiminnasta suoritettiin.

Analyysissä selvisi, että kehys paransi ennustuksen osumatarkuutta 32 % suhteessa yksittäiseen parhaaseen regressiomalliin.
Tuloksen tilastollinen merkittävyys vahvistettiin Wilcoxonin menetelmällä.
Täten johtopäätöksenä voidaan todeta, että kyseinen lähestymistapa on parempi kuin mikään käytetyistä yksittäisistä regressiomalleista.
ED:2014-10-05
INSSI record number: 49805
+ add basket
« previous | next »
INSSI