We start with exploring and munging data from datasets provided as part of the NASA COVID 19 challenge to get case and deaths of Vancouver, Canada, NYC, and Taipei, TW. We then come up with a couple of interesting hypotheses on why the infection/death rates across these three cities are so different and look at datasets and trends including:
We chose this project because there are a wealth of hypotheses out there but nigh three of us could find any data driven discussion/observation/conclusions.
We started by looking at available datasets then and brainstormed that against interesting questions we ourselves, personally, have about COVID19.
We started with space agency data and then enriched it (a "data pipeline" if you will :P). We used Azure Storage for storing most of our datasets (which included imagery fetched through Bing Image, News Search) and also ran them through pre-trained machine learning models that helped us detect, localize, and classify the count of masks (via Computer Vision on Azure) as well as analyze sentiment on common news articles from each city during each phase of the outbreak (using Azure Cognitive Services Sentiment Analysis).
This was super interesting as we actually used data sourced first from NASA to draw conclusions on some of the assumptions we already had about the above four questions!
Also under review - just need to double check a couple narratives to make sure they are OK to share as it uses Microsoft tech! Please contact sacha@microsoft.com, vimura@microsoft.com and Lindsay.Milliard@microsoft.com for more tomorrow in the AM!
Space Agency Provided Datasets
https://www.ncdc.noaa.gov/cdo-web/datasets/GHCND/stations/GHCND:USW00094728/detail
https://search.earthdata.nasa.gov/search/granules?p=C1288777589-LARC&g=G1622562964-LARC&m=0!0!1!1!0!0%2C2&tl=1575179710!4!!
Open Source Datasets
https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports
MSFT Azure AI/ML Services Used