From 2ce1fc194c5db0c31e9b0713767d5e9a84289aac Mon Sep 17 00:00:00 2001
From: Ryan Lagerquist <32083417+thunderhoser@users.noreply.github.com>
Date: Fri, 18 Nov 2022 12:24:11 -0700
Subject: [PATCH] Update README.md

---
 README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index bf9f43e..b32563a 100644
--- a/README.md
+++ b/README.md
@@ -9,5 +9,5 @@ This repository contains several Jupyter notebooks for implementing the methods
  - `classification_mc_dropout.ipynb` implements MC dropout for digit classification.
  - `classification_npdp.ipynb` implements quantile regression for digit classification, using a special NN architecture that prevents quantile-crossing.  Both notebooks use the spread-skill plot and discard test to evaluate uncertainty estimates.
  - `crps_loss.ipynb` implements the continuous ranked probability score (CRPS) as a loss function for ensemble prediction (EP).
- - `regression_multi_datasets.ipynb` implements three UQ approaches (PDP, EP, and MC dropout) for six synthetic datasets.  It creates four evaluation graphics to evaluate the models (attributes diagram, spread-skill plot, discard test, and PIT histogram).
- - 'regression_multi_model.ipynb' implements four UQ models (PDP_Norm, PDP_SHASH, EP_CRPS, and MC dropout) and compares them on a synthetic dataset (selected from six options).  It creates fout evaluation graphics (attributes diagram, spread-skill plot, discard test, and PIT histogram) and compares the models using eight different scores.
+ - `regression_multi_datasets.ipynb` allows the user to select one UQ approach (PDP, EP, or MC dropout), then compares the results across seven synthetic datasets.  It uses four evaluation graphics (the attributes diagram, spread-skill plot, discard test, and PIT histogram) and eight evaluation scores (MSESS, SSRAT, SSREL, MF, DI, PITD, CRPS, and IGN).
+ - `regression_multi_model.ipynb` allows the user to select one synthetic dataset, then compares the results across the three UQ approaches, using the same evaluation tools.