Comparing weather data with internet speed test data to determine if weather influences my internet speed
Comparing weather data with internet speed test data to determine if weather influences my internet speed
I wanted to put together a Jupyter Notebook of my analysis of whether outdoor weather conditions affected my internet speed.
I scrape weather data from Weather Underground, use internet speed data that I collected myself (using speedtest-cli on my Raspberry Pi), and then compare them to see if there’s any feature of the weather that predicts my internet speed.
Spoiler: there’s not.
I used linear regression because the data actually is fairly linear — it just has a slope of near-zero.
I also tried but didn’t include a K-Nearest Neighbors, Random Forest, and Gradient Boosting regressions, but those were pretty useless, too.
Ultimately, my data falls into the trap where the best accuracy I can get is to just predict that my speeds are always going to be good, because for a variety of reasons (not the least of which is that there’s an upper bound on my speed imposed by Comcast) weather isn’t a strong predictor of my internet speed.