Dataset can be found here.
First, let’s see just how many backers projects rising from each country gather.
Surprisingly, projects from the US don’t have most of the backers. Rather, the title goes for Hong Kong.
Whilst we’re on backers, let’s dissect the histogram of backers for successful projects with a “logical” goal.
As you can clearly see, the histogram is skewed to the left. Meaning projects with smaller amounts of backers are more prevalent.
Now let’s pit backers against categories.
Let’s jump from backers to something else. The following chart shows the days between goal date and start date of canceled projects.
I just have one stacked chart for this post, and that’s country-category-pledged-state.
I have another histogram for you, the logical goals.
Now let’s do pledged vs. state of the project.
Another histogram, this time, pledges. Another skewed histogram.
Finally, I have three scatterplots, two being regressions.
First, the one that isn’t a regression. It basically shows the real pledge value, as in, adjusted by inflation vs. the pledged value.
Now, I trained a Stochastic Gradient Descent Regressor on these data:
- Category - Backers -> Pledged
- Category - Goals -> Pledged
Here’s the first regression plot:
A regression with normal distribution. Fine. But the next one is just unmaintainable.
Well I hope this article was fun and informative.
About the Author: Chubak Bidpaa is a data programmer, specializing in the wide area of web spiders, machine learning, and deep learning. You can contact him on Discord with Chubak#7400 and email [email protected]