search

EHWUSF / HS68_2018_Project_1

0 stars 9 forks source link

Feature Proposal of Data Visualization #2

Open RoxanneXin opened 6 years ago

RoxanneXin commented 6 years ago

I'd like to do something about plots. There're countless of parameters in matplotlib we can try to figure out. For example, we can split multi predictors and plot relationships with the dependent variable for each predictor. It can help to show the relationship directly and explain the coefficient well.
Also create a theme and different color set will be convenient for fancy graphs and animations. Just a brief idea, please correct me or give me more suggestions.

And we can randomly test the correlation b/w independent variables to justify the use of linear regression models (thanks Rohit for adding this one).

RoxanneXin commented 6 years ago

Also residual, RSS and R^2 plots for best subset selection.

hhan14 commented 6 years ago

I like your idea about data visualization supporting understanding the relationship among data points. I think you can start gearing up from comparative analysis with other tools such as Gnuplot. I think the main caveat for matplotlib is that the tool is so dependent on Numpy, which is not compatible with other programs but only can used within Python.

NikitaThomas commented 6 years ago

I think this idea will go in hand with what I am working with. I can include your graphs as options for my prompt asking what type of plot the user would like to begin visualizing and then give more prompts from there depending on the type of plot. We can collaborate so that I can incorporate your functions with my code and vice versa.