1
$\begingroup$

https://www.kaggle.com/kemical/kickstarter-projects

I was checking the distribution of the "goal" feature in this 2016 KickStarter data set. I figure goal would be more normally distributed but it wasn't. I checked the Kick Starter website for published statistics. It seems the sampling was random as certain categories appear at the frequency Kick Starter Published but the data didn't meet the required assumptions for a T test. I check the Kaggle site for a summary of the goal data and the distrubtion looks the same but I'm very skeptical. I made a qq plot to check normality but this really can't be right.

What type of distribution is the column 'goal' in this data set? If the distribution should look like this why? Maybe I broke something or put in the wrong code and I wanted to be sure. My gut is telling me to take a serious look at the goal column.

enter image description here

Improve this question
$\endgroup$
2
  • $\begingroup$ To understand how a numerical variable is distributed, it's best to firstly plot it's histogram (plt.hist(variable)). From there, from the "shape" of the distribution, you can get an idea of how the variable could be distributed. If you need any suggestions on this, then if you could edit this question and include a histogram over the data, that would be amazing. $\endgroup$
    – shepan6
    Commented Jun 17, 2020 at 16:30
  • $\begingroup$ I did plot a histogram but the shape isn't what I expected it to be. My gut is telling something is wrong. Could you plot a histogram to confirm it matches with what kaggle provides? $\endgroup$
    – Laurent
    Commented Jun 17, 2020 at 18:22

1 Answer 1

Reset to default
0
$\begingroup$

Vaguely looks like it's pareto distributed, meaning that most goals are small, some are medium, and a few are REALLY big. It's fat-tailed and not normally distributed at all. This isn't too surprising- projects vary a lot in nature, there's no reason to expect them to be clustered around a particular value.

Caveat: There's bunching effects at certain integral number values.

A nice way to see it is to do a histogram plot with log=True and bump up the number of bins. enter image description here

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.