#### Checkout report.pdf for a summary of our work on predicting instagram virality. #### 

Profiles folder contains post data for 16539 images from 972 Instagram influencers.
Data for each profile is a JSON blob of the form:
	"alias",
	"username",
	"descriptionProfile",
	"urlProfile",
	"urlImgProfile",
	"website",
	"numberPosts",
	"numberFollowers",
	"numberFollowing",
	"private",
	"posts": a list of JSON blobs corresponding to each post.

Each post is blob of form:
	"url",
	"urlImage",- Note: in some of these this is a list
	"isVideo",
	 "multipleImage",
	"tags", - being hashtags
	"mentions",
	"description",
	"localization",
	"date",
	"numberLikes",
	"filename"

Thanks to gvsi for the orginal part of this dataset. We have built upon this dataset by implementing a webscrapper
for Instagram using Selenium. 

Note: Instagram regenerates image urls, so we had to write a script to regenerate stale URLs for the dataset of posts 
we took from gvsi. We also dropped posts that no longer exist, and files for users that no longer exist. The original dataset 
now consists of 15178 images.