Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- Data Viz project ideas:
- 1) Dynamic Infographic
- -The most popular way to present easily digestible data is an infographic, currently. Making it dynamic could make it
- interactive and interesting.
- -Curious if Pokemon data is readily available in csv format. Could have it such that I could sort Pokemon by type and
- have interactible stats for each Pokemon.
- -pros:
- -I love Pokemon and it's something that if I made it, it could be interesting to people beyond this
- class.
- -Pokemon is VERY data-heavy with many parameters (primary type, secondary type, six base stats, primary
- ability, secondary ability, Pokedex #, etc., can even get into what Pokemon can do which moves) so
- there's LOTS of room to play, especially since Pokemon is a highly statistical game
- -cons:
- -Doubtful that someone has made a comprehensive csv including all Pokemon stats.
- -Even if there was, I think making something presentable in something so popular would require
- more effort than I can exert in a few weeks' timeframe. Especially since I'm a perfectionist and would
- try and do too much at once and probably run into several bugs and maybe give up on the project.
- 2) Use machine learning in Python (ask professor if this is ok) or maybe even pj5s to generate a visualization and ask for an
- input to predict an output. Have the user click around to learn how it works. This would be a way to add user interface while
- implementing ML
- -get exoplanet data in csv form and have people test out various parameters like "distance from star" "type of star"
- "mass" and have them select which parameter they want to test to see what kind of planet would form ("hot jupiter"
- "terrestrial" "gas giant" etc.
- -pros:
- -interesting
- -data likely readily available en masse
- -cons:
- -would require me watching days of videos that I couldn't possibly complete a presentable assignment in
- the timeframe of the assignment
- - Use one of the existing databases specifically designed for machine learning.
- -pros:
- -easier to set up
- -cons:
- -I already browsed the database and there isn't anything super interesting in there. Plenty of cool
- things but teaching myself data-crunching and data-visualization through machine learning is already
- something fairly intense when I barely understand Python as it stands. I'm able to do it, but I can't
- fully grasp it at a fundamental level yet. The theory, yes. The implementation, not as much.
- 3) Use p5js or d3js to do live lookups of weather data (apparently there's a site called openweathermap that uses JavaScript to
- provide limited amounts of data for free) and present the information in an interesting manner.
- -Could limit it just to Orlando, or have it so that anyone could enter information and I'd build my own little weather
- applet.
- -pros:
- -I'm very knowledgable about the weather, so I'd be set on knowing how to interpret weather data
- in a human-understandable way.
- -cons:
- -I'm not at all familiar with how to process live data so this is probably a project I save for another
- day when I'm far better versed in JavaScript
- 4) You know those "digits of pi" visualization charts? Make that, but for any float. Would be more complex because I'd have to
- have it compute sqrt(x) or other predefined transcendentals up to 10,000 decimal points.
- -Visually aesthetic and more like real data visualization.
- -pros:
- -there's probably documentation on how to build a basic generator like this
- -cons:
- -might be too easy and not something worthy of a "final project" because it's something I could
- reasonably build in 3 days. Far too easy.
- 5) A fractal is a beautiful form of pure mathematical data visualization.
- -So you know how the Mandelbrot Set is defined to be z_n = (z_(n-1))^2 + z_0? There are variants of the Mandelbrot set
- that use different exponents, cubed, to the fourth, etc. that would be fairly easy to have user input to change (input
- a float, although due to the weird errors with float mixed with the chaos theory of fractals, it'd be safer to stick to
- int). Could also set it to Julia mode and have it generate fractals where z_0 is instead a c value.
- -pros:
- -Is sufficiently difficult but something I could realistically could do within the timeframe of the
- course
- -Has been a goal of mine to program for years
- -cons:
- -Doesn't quite satisfy the requirements of the project
- -Would be better on a GPU-enabled computer and I'm using a Surface Pro 2, which has decent CPU
- but a terrible GPU.
- 6) Facebook friends-of-friends graph nodes
- -There's a tight-knit group of friends I have, the objectum-sexual friends, who are friends with each other. Because
- there's a lot of drama, some friends aren't friends with others, so there is some disconnect with who is friends with whom.
- It's a perfect template to create a node graph of the friend network.
- -pros:
- -relatively easy to collect the data (I could sample 10 friends and their mutuals who are also in the
- community)
- -would be a fun one to set up.
- -cons:
- -This project might cause drama in the community and I lack the energy to deal with the drama.
- -Could create a fictional list of friends-of-friends.
- -pros:
- -will not cause drama with my actual friends (or maybe it will because some might wonder why I didn't
- use my actual friends list)
- -cons:
- -Rather boring and might be a little too easy.
- 7) Portal 1 has a mode that collects data on how many steps and portals you used for a particular test chamber.
- - The game takes about 1 hour to play through and I've already practically memorized the entire game, but haven't
- optimized my strategy. Play the tracking mode and collect step and portal count data 5-6 times (which is nothing) and make
- some sort of data viz chart of each test chamber and see what happens when I optimize for steps vs. when I optimize for
- portals.
- -pros:
- -I love Portal and it'd be fun
- -cons:
- -Won't be able to create anything impressive programming-wise; only something informative for the game
- 8) Don't even input any data. Generate a framework for a user to input 2D (or even 3D) data and have it plot to the screen.
- - Since I'm already familiar with 3D Processing, and am in the process of learning p5js, I could essentially have it so
- that with user input, it'd add parameters to points and plot them on the screen. Have it fully interactable. Import
- an actual library so that the navigation isn't crap like it was in the one I wanted to try building from scratch. Make
- it so that it can actually calculate regression lines, even in 3D.
- -pros:
- -a culmination of what and all I've been learning throughout the class
- -the right level of difficulty and the logical next step, maybe?
- -If I do this nicely, I can even use this to export custom graphs for actual coursework that requires
- data plotting.
- 9) Do something simple, BUT do it in as many different programming languages as possible
- - Take a basic data set, and try to represent it as a very simple bar graph or a scatter plot in as many programming
- languages as I can. Processing, pj5s, Python, C, C++, C#, etc.
- -pros:
- -Forces me to learn the basics of a variety of OOP languages which will be very helpful moving forward
- -Would be an interesting learning experience to see how I can do the same thing in different languages
- -cons:
- -Would require me watching days of videos just to get even one other programming language to the level
- I can plot charts.
- -Unsure of how helpful this would be in the future as I might get confused learning syntax of various
- languages in parallel, especially the more similar ones.
- 10) Take one data set, and try and represent it in as many different simple ways as possible in Processing, p5js, or d3js.
- -Take a basic data set and try and represent it in as many different visualizations as possible. Have these be selectable
- to the user, with a little pop up with information on the pros and cons of each graph.
- -pros:
- -Using what I already know in ways I didn't know I could do.
- -Learn how to create different data viz structures
- -Could be very useful as reference to other students if I get it done long before the project is
- actually due, and could format it so that future students get a ReadMe that explains how to
- interpret the code. With 10 different easy ways to visualize data and an explanation of each,
- it might help them with their own code and ideas of their own.
- -cons:
- -I'm not learning anything really useful for my long-term goal of machine learning
- -Sort of underwhelming in presentation (but still, if it helps other students, it might be worth my
- while creating several simple applets, even if it means not using classes to define data structures in
- one applet)
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement