Skip to content

Experimental analyses of LaMem score as a predictor of social media likes.

Notifications You must be signed in to change notification settings

xke/lamem-social-media

Repository files navigation

Image Memorability and Social Media Likes

I was intrigued by the news headline that "Scientists have uncovered exactly what makes a photo memorable" and wanted to better understand whether this LaMem algorithm might be able to predict the number of social media "likes" for a particular image.

Fortunately, it's relatively fast to use the Instagram API and the LaMem API to get output data that can be graphed to better understand the correlation between social media "likes" and the "memscore". Check out the Python code.

I have not found any strong correlation between image memorability (as defined by the LaMem algorithm) and # of social media likes.

Here's the output based on 70 data points from Taylor Swift (the most popular Instagram celebrity):

Linear Regression Image

  • X axis = Number of Instagram Likes

  • Y axis = MemScore from the LaMem algorithm

  • Correlation coefficient (r): 0.0933 (very low)

(Linear regression output via http://www.alcula.com/calculators/statistics/linear-regression/.)

About

Experimental analyses of LaMem score as a predictor of social media likes.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages