Supervised Machine Learning : Accuracy Comparison of Different Algorithms on Sample Data
Acharya, Tikaram (2021)
Acharya, Tikaram
2021
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-202102102161
https://urn.fi/URN:NBN:fi:amk-202102102161
Tiivistelmä
The purpose of the final year project was to discuss the theoretical concepts of the analytical features of supervised machine learning. The concepts were illustrated with sample data selected from the open-source platform. The primary objectives of the thesis include explaining the concept of machine learning, presenting supportive libraries, and demonstrating their application or utilization on real data. Finally, the aim was to use machine learning to determine whether a person would pay back their loan or not. The problem was taken as a classification problem.
To accomplish the goals, data was collected from an open data source community called Kaggle. The data was downloaded and processed with Python. The complete practical part or machine learning workflow was done in Python and Python’s libraries such as NumPy, Pandas, Matplotlib, Seaborn, and SciPy.
For the given problem on sample data, five predictive models were constructed and tested with machine learning with different techniques and combinations. The results were highly accurately matched with the classification problem after evaluating their f1-score, confusion matrix, and Jaccard similarity score. Indeed, it is clear that the project achieved the main defined objectives, and the necessary architecture has been set up to apply various analytical abilities into the project.
To accomplish the goals, data was collected from an open data source community called Kaggle. The data was downloaded and processed with Python. The complete practical part or machine learning workflow was done in Python and Python’s libraries such as NumPy, Pandas, Matplotlib, Seaborn, and SciPy.
For the given problem on sample data, five predictive models were constructed and tested with machine learning with different techniques and combinations. The results were highly accurately matched with the classification problem after evaluating their f1-score, confusion matrix, and Jaccard similarity score. Indeed, it is clear that the project achieved the main defined objectives, and the necessary architecture has been set up to apply various analytical abilities into the project.