https://github.com/briochemc/telluridefigure

Test hypothesis on DFe/FeT dependence on FeT
https://github.com/briochemc/telluridefigure

Last synced: 3 months ago
JSON representation

Test hypothesis on DFe/FeT dependence on FeT

• Host: GitHub
• URL: https://github.com/briochemc/telluridefigure
• Owner: briochemc
• Created: 2018-11-02T17:20:56.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-11-06T01:15:52.000Z (over 6 years ago)
• Last Synced: 2024-10-12T10:30:19.996Z (7 months ago)
• Language: Julia
• Size: 457 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: ReadMe.md

Awesome Lists containing this project

README

        
# Generating other versions of Figure 1 of report

In all cases the *R*2 values are quite low (< 0.5).

Hence, I believe that if the observational datasets yield signinficantly higher *R*2 values (e.g., > 0.8) for the same test, then the argument made in the report does not hold.

## Using uniform distributions

![image1a](https://github.com/briochemc/TellurideFigure/blob/master/dissolved_Fe_fraction_from_uniform_DFe_and_FeT.png)

Figure generated with uniform distributions of DFe and FeT (then removing samples with DFe > FeT).

## Using lognormal distributions

### lognormal DFe and FeT that matches the mean and variance of the uniformly generated DFe and FeT above

![image1b1](https://github.com/briochemc/TellurideFigure/blob/master/dissolved_Fe_fraction_from_lognormal_DFe_and_FeT.png)

Figure generated with lognormal distributions of DFe and FeT (then removing samples with DFe > FeT, but should not happen).

### lognormal DFe and FeT, but with a higher mean and variance for DFe

![image1b2](https://github.com/briochemc/TellurideFigure/blob/master/dissolved_Fe_fraction_from_lognormal_DFe_and_FeT2.png)

Figure generated with lognormal distributions of DFe and FeT (then removing samples with DFe > FeT).

### lognormal DFe and PFe (more natural?)

![image1c](https://github.com/briochemc/TellurideFigure/blob/master/dissolved_Fe_fraction_from_lognormal_DFe_and_PFe.png)

Figure generated with uniform distributions of DFe and PFe (and then FeT = DFe + PFe, no need to remove samples).

This last figure seems more natural to me because I believe generating DFe and FeT and then imposing DFe < FeT biases the sample distributions.