ETL processing using machine learning

Vladislav Shlapak

Bakalářská práce

Abstract:

The twenty-first century is the age of the information boom. Technology advances at a rapid pace, but so does the amount of information. A person today receives more information per day than our ancestors did in decades. The information must be stored and processed to obtain knowledge. Decision-making in many industries depends on the timely acquisition of knowledge from information. The faster and better the processing and analysis of information, for example, the more money it will bring to the business. Today, one of the most popular ways to store large amounts of data is the data warehouse concept. And the process of extract, transformation and loading is called the ETL process. This stage is the main one for getting started with information, but it is also the most time-consuming and error-prone. To increase the speed and accuracy of information processing, this work uses a deep machine learning model that will classify the type of trader's transaction by fields. I will explain why a deep learning model is needed to solve this problem and why automation of the ETL process is the next step in the development of a more reliable and up-to-minute system of data management.

Keywords

data science, mlps, classification, supervise, etl, deep learning, keras api, tensorflow, transactions, trading, python

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 2. 4. 2023

Obhajoba závěrečné práce

• Obhajoba proběhla 22. 5. 2023
• Vedoucí: Ing. Jan Pavlík, Ph.D.
• Oponent: Miloš Ulman, Ing.

Česká zemědělská univerzita v Praze

Provozně ekonomická fakulta

Bakalářský studijní program / specializace:
Environmental Data Science / Informatics