COMPARISON BETWEEN STOCHASTIC GRADIENT DESCENT AND SUPPORT VECTOR MACHINE ON ECOMMERCE DATA USING RANDOMIZEDSEARCHCV

ARAUNA, KEVIN VICTOR (2023) COMPARISON BETWEEN STOCHASTIC GRADIENT DESCENT AND SUPPORT VECTOR MACHINE ON ECOMMERCE DATA USING RANDOMIZEDSEARCHCV. Other thesis, Universitas Katholik Soegijapranata Semarang.

	Text 18.K1.0011-KEVIN VICTOR ARAUNA-COVER_a.pdf Download (1MB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-BAB I_a.pdf Download (118kB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-BAB II_a.pdf Restricted to Registered users only Download (249kB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-BAB III_a.pdf Download (230kB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-BAB IV_a.pdf Download (848kB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-BAB V_a.pdf Download (1MB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-BAB VI_a.pdf Download (113kB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-DAPUS_a.pdf Download (337kB)
	Text 18.K1.0011-KEVIN VICTOR ARAUNA-LAMP_a.pdf Download (1MB)

Abstract

In recent years, the use of e-commerce or online shops has increased considerably. Various online stores have sprung up on the internet, both small and large-scale. This has a very important effect on the effective use of time and the level of sales figures. With so many e-commerce emerging and even with almost the same type or product, it will be quite difficult to reach potential buyers for them. The intentions of consumers will also be different for each online store they visit. So the problem with the number of e-commerce that has sprung up is that shop owners must have data that is effective enough to see the intentions of consumers in their stores whether to buy their products or not, regardless of the marketing strategy they apply. Every e-commerce site has visitor history data especially nowadays that everything is online, it will make it easier for store owners to track consumers who visit their online store. This data includes all information regarding the traces of consumers who visit online stores, both new and repeated. With this data, processing will be carried out to make predictions about consumers' shopping intentions. The data contains attributes that will later be used for processing with the algorithm to be compared. Several algorithms will be used to process this data in order to find which one is the best in predicting consumers' intention to make a purchase or not. The comparison of algorithms will be carried out using the value of the Precision, Recall, F1-Score, Accuracy. Also some values that are considered to support the results of the comparison. The algorithm are Stochastic Gradient Descent (SGD) and Support Vector Machine (SVM). Comparison of the two algorithms is supported using RandomizedSearchCV as the hyperparameter tuning. The result of which algorithm is better with hyperparameter tuning is SGD. SGD obtained an accuracy value of 0.89 with 60% test data and 40% training data

Item Type:	Thesis (Other)
Subjects:	000 Computer Science, Information and General Works > 004 Data processing & computer science
Divisions:	Faculty of Computer Science > Department of Informatics Engineering
Depositing User:	mr AM. Pudja Adjie Sudoso
Date Deposited:	04 Apr 2023 06:03
Last Modified:	04 Apr 2023 06:03
URI:	http://repository.unika.ac.id/id/eprint/31400

Actions (login required)

View Item