Join MathWorks and FactSet for a joint seminar discussing FactSet’s Data Model for analysing content and the application of Deep Learning & Natural Language Processing techniques in building a sentiment-based stock selection model from this content.
FactSet’s Data Model for Analysing Structured and Unstructured Content
Chris Thomas, Vice President, Product Strategist, FactSet
In the constantly evolving world of financial big data, the largest issue facing analysts is not the availability of unique content to drive alpha, but the usability of that data. Finance does not have a big data problem, it has a bad data problem. FactSet has come up with a sophisticated entity driven data model that allows analysts to seamlessly connect across data sets and spend less time scrubbing data and more time analyzing it. Now, FactSet is taking its 40 years of industry expertise and applying it to the world unstructured data to allow our users to look beyond traditional structured content sets and turn information into intelligence.
Sentiment Analysis of FactSet Transcript History using Machine Learning & Deep Learning with MATLAB
David Willingham, Senior Data Analytics Engineer, MathWorks
This session will highlight how to build a custom sentiment scoring model of FactSet’s Earning transcripts using Natural Language Processing (NLP), classical Machine Learning and Deep Learning functionality in MATLAB. The sentiment is analysed, scored and compared with intraday prices, giving an automated & scalable process for assessing companies against their stock market performance.
We then highlight how the final model can be integrated & deployed into a custom web dashboard for users to interact with.
Registration is free, however places are limited. Drinks and canapés will be served at the end of the seminar.