Using ML for GDP Prediction- ON-502

Project type: Research
Desired discipline(s): Computer science, Mathematical Sciences, Mathematics, Economics, Social Sciences & Humanities
Company: MacroXStudio
Project Length: 6 months to 1 year
Preferred start date: 09/06/2021
Language requirement: English
Location(s): Toronto, ON, Canada
No. of positions: 2-3
Desired education level: Master'sPhDPostdoctoral fellow
Search across Mitacs’ international networks - check this box if you’d also like to receive profiles of researchers based outside of Canada: 

About the company: 

We are an early stage fintech startup based in Silicon Valley with a subsidiary in Toronto. Our mantra is to use Big Data and AI to “Know Faster and Invest Better.” Better returns are not just more retirement income for our investors but also meaningful social change.  Specifically, we are  measuring the world macroeconomy faster and better using thousands of new data sources like twitter, search, credit card, satellite data etc., rather than the incomplete, months delayed, slow, and frequently revised Government data. For example, even many advanced countries are struggling to quantify the impact of COVID-19 on unemployment rate and countries like India are using private think tanks to measure unemployment since there is no government data. You can help MacroX change this!

In addition to providing superior returns to institutional clients, we will provide free access to our cleaned data to relevant supra-national bodies like the UN to help make better social decisions. We have led teams at MSFT, Wall Street, and have had a Unicorn exit. You will also get exposed to other top experts such as faculty at HBS, World Bank leaders etc. Here is a talk on applying AI to Asset mgmt.

Describe the project.: 

Based on the pioneering research of prominent Harvard professors and the CEO, MacroXStudio would like to use Machine Learning to predict countries’ GDP. The students will leverage the existing research paper and available code to create a proprietary automated ETLV (extract-transform-load-visualize) modular pipeline to predict a countries GDP.  The goal of this project is to produce an end to end predictive model pipeline using novel machine learning and data science techniques. MacroXStudio will provide the data access, the initial code, the templates, and expert guidance.

The Deliverables of The Project:

  • Optimize and Validate existing predictive Model.
  • Create an end-to-end automated ML pipeline for the deployment of the model.
  • Clear documentation of the Data Analysis and Machine Learning WorkFlow decisions made throughout the project’s life cycle.

Required expertise/skills: 

Intern Degree level = Masters, PhD, or Post Doctoral Fellow
Strong understanding of:

o   Statistics

o   Big Data

o   Data Exploration

o   Feature Generation

o   Training Machine Learning Models in Python and R. You also need at least 3 graduate level courses in statistical learning.

o   Deployment of Autonomous Machine Learning Pipeline (Cloud Deployment)

o   Modular Programming

Ability to adapt open source code for the purposes of the project.
Strong writing skills to communicate and document processes and findings