An attempt at quantifying New Year’s Resolutions

This new year I wanted to try to track how well people followed through with their New Year’s Resolutions. I attempted to do the same last year by pulling the number of certain hashtags from Instagram but I wanted to expand on that.

I’m currently in the process of going through a 30-day yoga challenge on YouTube. In the middle of one of the videos I thought, why not look at the number of views throughout the 30 days? Since many of these series are posted on New Years Day or people start them on New Years Day, it might be a good indicator of how many people follow through with their resolutions. Even if they are started or posted at a different time of year, I’d wager they follow the same trend since it’s really just tracking how well people stick to new habits.

So, I began to record the number of views of a few 30 day challenges I found on YouTube. All of the ones I found were yoga/fitness related. When picking them out, I made sure each series had a good amount of views, had all 30 days of videos (I did include one that had a few missing), and was specifically made to be a 30-day challenge (not just 30 random videos cobbled together). I ended up finding 8 different series to use.

channel 1 2 3 ... 28 29 30
0 adrienne rev yoga 911531 545313 429422 ... 68497 66419 62551
1 adrienne yoga 8910198 3850409 2252723 ... 428423 404578 419763
2 adrienne camp 1795098 901061 651103 ... 150629 157835 158397
3 pilates 80209 46725 37091 ... 10266 8573 10069
4 hiit 95189 42773 50805 ... 11964 43869 17306
5 weight loss 289484 98556 52952 ... 18924 6795 14210
6 do you yoga 53792 23386 16617 ... 2294 2576 3272
7 fightmaster 291284 108261 73509 ... 13391 12774 16571

To provide a fair comparison we’ll have to normalize the data to the first value. This way we can compare the percentage of people continuing at each day. Plotting the results we get the following scatter plot and box plot:


It seems like they all follow the same trend with a pretty steep drop off.

I mentioned in the beginning I tried doing something similar with Instagram. I saw a few friends were doing the 365 photo a day challenge. The goal is to post a photo each day usually using a hashtag like #day42. So I scraped the instagram website to get the total counts for all hashtags from #day1 to #day365.

There are two scatter plots below of the data. One that that shows the number of hashtags throughout the whole year and another showing the same info but just for the first 30 days of the year.

Day 1 2 3 ... 363 364 365
hashtags 4096062 3600835 2584125 ... 3119 4175 11866


Looking at the plot for the whole year there’s a few strange bumps. I’m guessing maybe this might be because there are other X day challenges due to the drops at 30 and 100. This probably means our data from Instagram is a bit more noisy then the data from Youtube. Plus there’s probably other reasons someone might use a hashtag like #day42 besides some challenge.

Even with that being said there still seems to be trend worth comparing especially from what we can see in the 30 day scatter plot. So let’s normalize it like we did before and compare it to our YouTube data. The data from Instagram is in red.


The data from Instagram seem follow the same trend, perhaps more on the forgiving side of things. But overall it seems most people don’t follow through with their New Year’s Resolutions (can’t say I’m too surprised). From our data, it appears only about 10% of people are still sticking to them after 30 days.