Decision Tree: Gini
Accuracy achieved: 50%.
Financial Data Forecaster
by Sumer Malhotra
Aim
This project aims at comparatively developing and testing various ML classification algorithms on stock prices. The project accepts a list of tickers and time range as input and outputs 1 or 0 for each stock; 1 if it's going up or 0 if it's going down.
Objectives
Some factors the algorithm will take into the consideration include various financial indicators like Price-to-Earnings, Price-to-Book, Debt-to-Equity, etc. along with various return statistics and Asset sentiment.
Supervisor
Name: Dr. Yip Chi Lap [Beta]
Email: clyip@cs.hku.hk
Telephone: +852 28578431
Office: Room 113, Main Building
Website: https://i.cs.hku.hk/~clyip/
Student
Name: Sumer Malhotra
Degree: Bachelor of Engineering in Engineering Science
Major: Computing & Data Analytics; Minor: Finance
Email: sumermalhotra1998@gmail.com
Telephone: +852 67486785
Method
Data
Collect time series financial data from various sources.
Statistics
Statistical analysis with pandas, numpy and scikit-learn.
NLP
Sentiment analysis with NLTK Vader and TextBlob.
Machine learning
Machine learning using scikit-learn.
Optimization
Continously working towards improving accuracy.
Results on Testing Set
Decision Tree: Entropy
Accuracy achieved: 50%.
Random Forest
Accuracy achieved: 70%.
Support Vector Machine: Linear
Accuracy achieved: 50%.
Support Vector Machine: Radial
Accuracy achieved: 60%.
Timeline
Detailed project plan and website
September 29, 2019
Literature review
October-November 2019
Data collection and preperation
December 2019
Algorithm development and testing
January-February 2020
Program Completion
March 2019
Final Report
May 2019
Poster
May 2019
