hyperparams screen | Data_1_Design_1_vs_Data_1_Design_2 | Omics-Datatype: PTX | Date-Time: 26_09_2024-14_27_02

Hotkeys

Press:
t --> Jump to Table of Contents and save current scroll position 📑
s --> Save current scroll position 📌
b --> Jump back to saved position 🔙
d --> Download all embedded files as zip 📥
e --> Write an email to contact info ✉

Report Info ℹ

omics data type :	PTX
data description :	Proteomics data
data collection date :	September 2024
meta condition :	Phase
meta batch :	Reactor,
limma design :	NA
analyst name :	Thomas Rauter
contact info :	thomas.rauter@plus.ac.at
project name :	DGTX
method description :	NA
results summary :	NA
conclusions :	NA

Downloads 📥

Venn-Heatmap
Venn-Heatmap long
Nr. Hits Barplots
Spline Curves

Venn-Heatmap

The 'Venn-Heatmap' is inspired from a Venn-diagram, in the sense two results are shown in distinct colors and the overlaps between the results are shown with the mix of the two colors. Every row in the 'Venn-Heatmap' is a feature that is a hit for at least one combination of hyperparameters. Every column is a combination of the 'internal' hyperparameters. By looking at this, one can see how many and which hits each 'internal' hyperparameter combo delivered for the two combinations of 'outer' hyperparameters.

Venn-Heatmap long

This is the same venn-heatmap as above, but in a format long enough to see all details.

Nr. Hits Barplots

The barplots show how many hits each 'internal' hyperparameter combo delivered for the two combinations of 'outer' hyperparameters.

Spline Curves

In this section, the spline curves with the data points for maximally 6 hits and 6 random non-significant features are shown for each combo of hyperparameters. This mainly allows to identify the spline parameters that lead to nice fits (not overfitting == curves too wiggly) and not underfitting (curves not matching pattern complexity)