Using AI / machine learning to solve real world problems

The field of Artificial Intelligence and Machine learning has made great advancements over the past few decades and has become more intertwined into the daily lives of people. With the development of technology, it is common to see machine learning methods used and adopted to help solve real-worl...

Full description

Saved in:
Bibliographic Details
Main Author: Lok, Ignatius Zhengrong
Other Authors: Ng Wee Keong
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2021
Subjects:
Online Access:https://hdl.handle.net/10356/148035
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:The field of Artificial Intelligence and Machine learning has made great advancements over the past few decades and has become more intertwined into the daily lives of people. With the development of technology, it is common to see machine learning methods used and adopted to help solve real-world problems of both individuals and well as large corporations. This project studies how different machine learning algorithms can be used to aid companies make better business decisions and to optimize their investments. In this particular project, we look to help TFI decide on new restaurant investments and potential locations. This is done by predicting the expected annual revenues of Turkish Restaurant based on the data given. The data provided is very imbalanced with a significantly larger test data compared to the training data. Additionally, the data provided contained obfuscated variables which encoded different categorical data types including Demographic, Real Estate and Commercial Data. We look to address the obfuscated data and the small training set during the preprocessing and feature engineering stage. Different supervised machine learning models including Random Forests, Support Vector Machines, XGBoost, LGBM and Ensemble learning methods are then applied to predict the restaurant revenue, allowing a better decision to be made when opening new restaurants and to increase the effectiveness of investments in new restaurant sites.