Mutual Fund Rating Prediction using Proportional Odds Logistic Regression with Imbalanced Class
Ashoka Wilson Dsouza1, Ismail B2
1Ashoka Wilson Dsouza, Department of PG Studies and Research Statistics, Mangalore University, Konaje (Karnataka), India.
2Dr. Ismail B, Professor, Statistics, Statistical Unit, Yenepoya Deemed to be University, Derlakatte, Mangalore (Karnataka), India.
Manuscript received on 13 February 2020 | Revised Manuscript received on 20 February 2020 | Manuscript Published on 28 February 2020 | PP: 6-10 | Volume-8 Issue-5S February 2020 | Retrieval Number: E10020285S20/2020©BEIESP | DOI: 10.35940/ijrte.E1002.0285S20
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Mutual funds ratings given by rating agencies, are very popular and helps new/first time investors to select and invest in funds based on the ratings a fund takes without going through the detailed portfolio. However sometimes these ratings could be biased or incorrect or in favor of specific fund and it could affect an investor decision. New investors face a lot of problems while investingand choosing mutual funds due to poor professional advice and lack of right tools and resources to assess a funds true performance. To overcome the problem of incorrect rating and to help an investor to choose the funds wisely using machine learning, we have attempted to predict the rating and classify mutual funds using proportional odds logistic regression which classifies funds intorating classes from 1 to 5 with 5 being the high rated fund and 1 being the low rated fund. While some prior studies have suggested methods of using clustering to classify based on performances using Supervised/Unsupervised learning, this paper deals with supervised learning forpredicting the ratings using the mutual fund financial ratios and also handles imbalanced classes.To handle imbalance class problem in a multi-class setting, we propose a new class balancing hybrid methodology of using EM and Gauss-Smote sampling that significantly improves the rating prediction.
Keywords: Classification, Gauss-Smote, Imbalanced Classes, Mutual Fund Rating, Proportional Odds Model.
Scope of the Article: Regression and Prediction