Analyzing the Hackaday Prize 2015

To coincide with Hackaday releasing today's Semifinalist and Best Product Finalist lists, we've created a new spreadsheet which should include all of the finalists:

2015 Hackaday Prize Semifinalist and Best Product Finalist Spreadsheet

Congratulations!

Over the last two days, we created a list of all of the projects tagged with the correct tags for the 2015 Hackaday Prize or Best Product*. You can find our spreadsheet here:

( Editor: This is now archived. 2015 Hackaday Prize and Best Product Spreadsheet.)

*(Note we're estimating if projects qualify based on the tags and our scraping - our columns on Best Product and the Quarterfinals are not canonical. Please let us know if you see an entry mislabeled.)

Projects Added Per Day:

Data - https://gist.github.com/anonymous/747e8acd2fae96421a93

Total Number of Projects:

Data - https://gist.github.com/anonymous/00ff97cfc206f78afe47

The data was obtained from the "Created on" project date and the graphs were scaled to the duration of the prize (so projects added before this aren't shown in the first graph).

While there is a small inflection at the end, the growth rate was pretty steady this year. Can anyone name the spike events?

And something different:

And if you hate angles:

Generated using tagxedo.

Word list here - https://gist.github.com/alpha-ninja/64a99368d230a22f2af2

A quick and dirty Top 100 by copy/paste the projects webpage text, then arrange the data in a spreadsheet pivot table

Grand Totals
Authors: 672
Projects: 910
Comments: 3 169
Skulls: 7 947
Followers: 22 421
View Count: 1 544 197

Top 100 projects and authors, by skulls, followers, views or comments

• Top 100 Projects by Skulls
• Top 100 Projects by Followers
• Top 100 Projects by View Count
• Top 100 Projects by Comments
• Top 100 Authors by Projects
• Top 100 Authors by Skulls
• Top 100 Authors by Followers
• Top 100 Authors by Comments
• Top 100 Authors by Views

Raw data available in the initial project, Statistics for 2015 Hackaday Prize Entries.

Have a *NIX based system? I've got some bash magic here to help you examine the Hackaday Prize 2015 entries, thanks to @alpha_ninja dumping all of the data as of 8/17/2015 into Github.

Yes, there are faster ways to do these things - but it's 2015 and these commands are fast enough for our purposes:

git clone https://github.com/alpha-ninja/had-data.git
cd had-data/had

# First, get all L.txt files, (which is the project log dump)
ls *L.txt | tr '\n' '\n' > filenames

# Quick cut - how many projects had > 3 logs?

for fn in `cat filenames`; do filelogs=$(grep -i "#discussion-list" $fn | grep -i "log" | wc -l); if [ $filelogs -gt 3 ] ; then echo $fn ; fi; done | wc -l
# Blecky, alpha_ninja and I all get 406

# Now do video references > 0 and logs > 3

for fn in `cat filenames`; do filelogs=$(grep -i "#discussion-list" $fn | grep -i "log" | wc -l); videos=$(grep -Ei -e youtube -e youkou -e video $fn | wc -l); if [ $filelogs -gt 3 ] && [ $videos -gt 0 ] ; then echo $fn ; fi; done | wc -l

# (Remove wc -l to see the file names)
# I got 219, which is lower than @Blecky and @alpha_ninja; use their lists most likely.

# Now, best product.  It is self-selected, so put everything in a file.

grep -i "bestproduct" * | grep "submissions" | awk -F'.' '{print $1}' > bproduct

# Now loop and search for greater than 7 logs and greater than one references to video

for fn in `cat bproduct`;  do fn2=L.txt; fnold=$fn; fn=$fn$fn2;  videos=$(grep -Ei -e youtube -e youkou -e video -e youtu $fn | wc -l); filelogs=$(grep -i "#discussion-list" $fn | grep -i "log" | wc -l); if [ $videos -gt 1 ] && [ $filelogs -gt 7 ];  then echo $fnold ;  fi;  done | wc -l

# I get 20 - check my work

Definitely check my work, and remove the "wc -l" if you want to see individual IDs (to check if it matches up with projects you know qualified, for example).

Have fun!

I took the most recent CSV (which does not include the most recent data on views), removed some projects with irregular data, removed some attributes that I did not want in the analysis, and uploaded the data into a S3 bucket on Amazon Web Services.

I then created an Amazon Machine Learning model with the modified CSV. Of the attributes included in the analysis, I decided that I most wanted to analyze the factors to determine the likelihood of not just winning a prize, but winning multiple prizes. So I set the "Prize Count" attribute as the Target. Below is a summary of the stats for the numerical categories:

My conclusion from the summary above is that Project Views will most likely translate to prizes on Hackaday!

...but wait, the model also included a categorical attribute:

Wow, the "Win Date" category (which is based on the first date with which a specific project won a prize) had the highest correlation towards winning multiple prizes.

In conclusion, if you wanted to win prizes in the first 10 weeks of giveaways, your best bet was to submit your project as early as possible, and update project logs as much as possible so as to garner more views from the "trending projects" feed. Skulls matter, but not as much as "Views" and "Date Submitted". Comments don't really seem to matter.

Going forward, I am not sure how reliable this data will be as prizes will now be based on community voting as well as inclusion in sponsor lists. All I know is that I am glad that I found Hackaday as a venue for my projects. Prizes are the best!

(I will edit this log with a link to my modified CSV shortly)

(Spoiler Alert: This may contain the future finalists)

This was easy to put together from the last log. The views:skulls suggestion was not as helpful as expected, but I think the last list "Top 10 Number of Times in Top 10 Lists" is a good indicator of whats happening here.

The Unofficial Top 10 Lists of the Hackaday Prize Prizewinners So Far:

A new and improved spreadsheet is now sorted by project, instead of by contest, making it easier to sort and analyze (no more double entries!) Get the spreadsheet, it looks much better. Download the new one here. It has presorted sheets for most Skulls, most Followers, and the new column sorted by a Views to Skulls ratio, with the data that was available (see feedback 6-10-2015.)
Someone won five times. You can find other information hidden in here also.

Column Names:
Project Name, Views, Comments, Followers, Skulls, Number of Entries, Data collection date, Number of Wins, Win Dates, and Views:Skulls.

Please check for mistakes. Thank you!

I finally finished the number of entries column, so "most" of the data is from 6-8.

Update: I added that stubborn LibreOffice pie chart at the end ;)

What does it take to be a winner? Getting an entry in early sure helps.
I think projects with pictures, videos, and clear details, look like they get more attention.

Here is the competition, see if you can find a pattern. Congratulations to all you winners!

The unofficial list of winners projects with views, comments, followers, skulls, and some with number of projects entered by the same hacker: