The Use of Twitter and Other Micro-Blogs are Examined to Determine if Consumer Sentiment Trends Can Assist Stock Analysts
Alex Isken in the AIM Room |
The seniors in the AIM program are beginning the semester learning about how to expand their fundamental analysis skills by studying 'big data analytics'. This week they have had the opportunity to learn from Dr. Terence Ow (Marquette University, Management Professor) and Alex Isken (AIM 2015 alumnus) about how to access and scrape big data from social media sources, clean and array the data, and conduct sentiment analysis.
Also assisting with the presentation was Dr. Ow's graduate assistant, YiJie Wang. He wrote most of the codes used in Python and R (the statistical analysis). YiJie was an important member of the team that prepared the workshop material.
Also assisting with the presentation was Dr. Ow's graduate assistant, YiJie Wang. He wrote most of the codes used in Python and R (the statistical analysis). YiJie was an important member of the team that prepared the workshop material.
We know that various data sources (i.e. Google Trends (https://www.google.com/trends/);
Google Analytics (http://www.google.com/analytics/)
and IBM Watson Trend (https://ibmwatsontrend.com/) have been used to monitor macro
economic and product trends. Dr. Ow and Alex Isken have shown this week that we can also dive into other social media and micro-blog databases (i.e. Twitter) to
extract raw texts and analyze them for individual company trends.
It has been suggested that it is possible to investigate real-time measurements
of collective mood states derived from large-scale Twitter feeds to predict the
sentiment about a firm’s products or services . This week the AIM students learned first hand how it is possible to use real-time social media
feeds to identify trends. They looked at Starbucks in an attempt to determine if various sentiment indicators are correlated with future revenues and the firm's stock price
movements.
Alex Isken and Dr. Ow in AIM 'big data' workshop |
The AIM students have been given an assignment this semester to conduct their own big data research. They are required to analyze the text
content of Twitter feeds over a specific time period to evaluate several sentiment (mood and polarity)
tracking tools. They will measure positive (+1) vs. neutral (0) vs. negative (-1) mood
and emotional states (i.e. fear, joy, sadness, anger, disgust, surprise,
happiness, etc.) and to quantify the sentiment the market holds for individual consumer products and services.
While their research will only scratch the surface of using real-time
data, they will learn the basics of coding (using Python), wrangling and arraying data (using Google Refine) and conduct a sentiment analysis (using R statistical software). This assignment will add another potential fundamental analysis tool to their set of skills.
Dr. Ow, Alex Isken and Dr. Krause |
Dr. Krause, AIM director, believes that this is the next level in the evolution of fundamental analysis and that graduates of the AIM program should be comfortable using this method of analysis.
He stated, "We appreciate Dr. Ow and Alex taking time out of their schedule this week to spend time in the classroom with the AIM students. We believe that this type of data analytics will continue to growth in importance across many disciplines. It is with workshops and assignments like this that our students are able to continue to refine their research and analysis expertise. Does Twitter and other micro-blog sentiment trends serve as a
predictor of a company’s future relative stock price? We'll see - and when we know - we'll probably not tell anyone!"