Upotreba obrade prirodnog jezika za predviđanje tržišta dionica

master's thesis

• Title (english): The Use of Natural Language Processing for Stock Market Prediction
• Author: Karlo Puh
• Mentor: Marina Bagić Babac (mentor)
• Committee member: Marina Bagić Babac (predsjednik povjerenstva)
• Committee member: Mirjana Domazet-Lošo (član povjerenstva)
• Committee member: Marko Horvat (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-06, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Predviđanje cijena na burzi oduvijek je bila zanimljiva tema budući da je usko povezana sa zarađivanjem novca. Korištenje obrade prirodnog jezika otvorilo je nove perspektive za rješavanje ovog zadatka. U ovom su radu kao mjerilo implementirani konvencionalni statistički modeli za predviđanje vremenskih serija. Zatim, radi metodološke usporedbe, implementirani su i testirani različiti najsuvremeniji modeli obrade prirodnog jezika u rasponu od osnovnih CNN-a i RNN-a do najnaprednijih BERT modela. Eksperimentalni rezultati pokazuju da postoji korelacija između tekstualnih informacija u naslovima vijesti i predviđanja cijene dionica. Model temeljen na GRU ćeliji s jednim linearnim slojem koji uzima parove povijesnih cijena i ocjene sentimenta izračunate pomoću BERT-a postigao je najbolje rezultate.

Abstract (english)

Predicting the stock market’s prices has always been an interesting topic since its closely related to making money. The use of natural language processing has opened new perspectives for solving this task. In this paper, the conventional statistical models for time-series prediction are implemented as a benchmark. Then, for methodological comparison, various state-of-the-art natural language models ranging from the baseline CNN and RNN to the most advanced BERT models are implemented and tested. Experimental results show that there is a correlation between the textual information in the news headlines and stock price prediction. The model based on the GRU cell with one linear layer which takes pairs of the historical prices and the sentiment score calculated using BERT achieved the best results.

• Keywords: predviđanje cijena na burzi; strojno učenje; analiza vremenskih serija; povratna neuronska mreža; BERT
• Keywords (english): stock market prediction; machine learning; time-series analysis; recurrent neural network; BERT
• Language: english
• URN:NBN: urn:nbn:hr:168:714750
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 15:57:53