Repository of colleges and higher education institutions

Show document
A+ | A- | Help | SLO | ENG

Title:Primerjalna študija metod strojnega učenja za modeliranje QSAR : magistrska naloga
Authors:ID Krivec, Matic (Author)
ID Panov, Panče (Mentor) More about this mentor... New window
Files:.pdf MAG_2024_Matic_Krivec.pdf (3,30 MB)
MD5: 3649D15B74B9DC0E364611C3BF14660E
 
Language:Slovenian
Work type:Master's thesis/paper
Typology:2.09 - Master's Thesis
Organization:FIŠ - Faculty of Information Studies in Novo mesto
Abstract:Magistrska naloga predstavlja primerjalno analizo desetih različnih QSAR-algoritmov strojnega učenja, uporabljenih na desetih različnih QSAR-podatkovnih množicah, ob uporabi različnih molekulskih odtisov, da bi ocenili njihovo učinkovitost pri različnih značilnostih in velikostih podatkov. Glavni cilj je bil oceniti, kako izbira algoritma vpliva na natančnost in robustnost modelov pri napovedovanju biološke aktivnosti. Vsak algoritem je bil sistematično testiran z uporabo metrik, kot sta korelacijski koeficient (R²) in koren srednje kvadratne napake (RMSE) za ocenjevanje uspešnosti algoritmov strojnega učenja. Rezultati kažejo znatne razlike v delovanju modelov, kar poudarja, da značilnosti in velikost podatkovne množice ter izbira prstnih odtisov kritično vplivajo na napovedno uspešnost QSAR-modelov. Ta študija ponuja vpoglede v izbire algoritmov za raznolike aplikacije v odkrivanju zdravil ter potrjuje pomembnost značilnosti podatkovnih množic pri QSAR-modeliranju.
Keywords:QSAR, strojno učenje, molekulski odtisi, vrednotenje modelov, napovedno modeliranje
Place of publishing:Novo mesto
Place of performance:Novo mesto
Publisher:M. Krivec
Year of publishing:2024
Year of performance:2024
Number of pages:XV, 106 str.
PID:20.500.12556/ReVIS-11349 New window
COBISS.SI-ID:223151619 New window
UDC:004.85(043.2)
Note:Na ov.: Magistrska naloga : študijskega programa druge stopnje;
Publication date in ReVIS:20.01.2025
Views:61
Downloads:1
Metadata:XML DC-XML DC-RDF
:
Copy citation
  
Share:Bookmark and Share


Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Link:http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.

Secondary language

Language:English
Abstract:This thesis presents a comparative analysis of ten different QSAR machine learning algorithms applied to ten diverse QSAR datasets, utilizing various molecular fingerprints, to evaluate their performance across varying data characteristics and sizes. The primary objective was to assess how the choice of algorithm influences model accuracy and robustness in predicting biological activity. Each algorithm was systematically tested using metrics such as correlation coeficient (R²) and root mean square error (RMSE) to measure predictive performance. The results demonstrate significant variations in model performance, highlighting that both the dataset's features and size, as well as the choice of the fingerprint, critically impact the efficacy of QSAR models. This study provides insights into algorithm selection for diverse applications in drug discovery and reinforces the importance of dataset characteristics in QSAR modeling.
Keywords:QSAR, machine learning, molecular fingerprints, model evaluation, predictive modeling


Back