# SNOTEL and CDEC ground\_measures - train vs. test
**URL:** https://community.drivendata.org/t/snotel-and-cdec-ground-measures-train-vs-test/6940
**Category:** Snowcast Showdown
**Created:** [December 25, 2021, 7:57pm UTC](https://community.drivendata.org/t/snotel-and-cdec-ground-measures-train-vs-test/6940 "2021-12-25T19:57:24Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@rszostak](https://community.drivendata.org/u/rszostak)
#### Post date: [December 25, 2021, 7:57pm UTC](https://community.drivendata.org/t/snotel-and-cdec-ground-measures-train-vs-test/6940/1 "2021-12-25T19:57:24Z")
What is the purpose of splitting SNOTEL and CDEC data into a training and testing set? Does this mean that ground\_measures\_test\_features.csv data cannot be used as input to predict SWE in cells from the submission\_format.csv file?
Thank you
---
### Author:  [@glipstein](https://community.drivendata.org/u/glipstein)
#### Post date: [December 27, 2021, 8:49pm UTC](https://community.drivendata.org/t/snotel-and-cdec-ground-measures-train-vs-test/6940/2 "2021-12-27T20:49:55Z")
Hi @rszostak - Thanks for reaching out about this. You may want to check out previous forum threads like [this one](https://community.drivendata.org/t/problem-clarification/6809/5) which are relevant to your question. In short, training data is provided to develop your solution, and test data is provided to assess its performance on a new set of data. The sites provided in the ground measures features CSV are distinct from the sites used for evaluation in the submission format. So you may use ground\_measures\_test\_features.csv as input in generating predictions for the submission format.
For more information you can check the description of pre-approved data sources [here](https://www.drivendata.org/competitions/86/competition-reclamation-snow-water-dev/page/417/#features).
Hope that helps!