Skip to content

Instantly share code, notes, and snippets.

View plorch's full-sized avatar
🏠
Working from home

Pat Lorch plorch

🏠
Working from home
View GitHub Profile
this is to see if i can geet gists to work
@plorch
plorch / Test
Created May 31, 2017 13:48
Test
## testing gist connection
- one
- two three
> Written with [StackEdit](https://stackedit.io/).
@plorch
plorch / gist:d975f0c084b44777ac57130947823625
Last active May 31, 2018 20:23
transfer: things that need doing
## Documentation
* add details on use to README.md
## Training for first testing of model
* Set up AWS E2 site for training images
* Train models for
* Species and empty
* Animal vs. empty
@plorch
plorch / AdamSmithBrandonEdwards_PredictiveModeling.md
Last active December 3, 2020 21:56
Road2Recovery2 2020 meeting to establish best practices to stop bird population declines. Each file is either notes from one talk or some daily notes

Adam Smith & Brandon Edwards – A strategy for integrating observational monitoring data and predictive modeling of population change

  • Trying to bring what we have learned from waterfowl to land birds
  • What to do about BBS gaps
    • Other observer-based surveys
    • Sound surveys
  • How to bring in other survey data?
  • Count = Density * Detectability offset
    • Detectability offset using qpad
      • p = prob of bird making a call
  • time and space will affect this
@plorch
plorch / AWSignite2020_CarlSverre_OperationalAnalytics.md
Created December 1, 2020 18:51
Notes from conference presentations

Operational BI: Executing ML models at speed and scale Carl Sverre from SingleStore https://virtual.awsevents.com/media/1_6hvzb0vo

  • share with Steve, India, and Adam
  • Operational analytics means using ML to train a model that can then be run as a function in a database
  • SingleStore provides a ml2udf function that allows you to create sql to implement the model in the db as a function
Poster sessions will have 2 min lightning talks
Exhibitors will have breakout rooms in daily sessions
From Timothy Maguire to Everyone: 01:14 PM
https://github.com/woobe/IAGLR_2021_H2O_Workshop
* Auto ML does model choice and tuning
* Why should I trust the model? Explaining the model is important
* Explainability toolkit
* Details
* package contains a split_frame function to get training and test
* can trian on multiple folds to use different subsets to train
* Lots of models are available
A Next Generation Research Database to Harness Great Lakes Environmental 'Omics Data Gregory J Dick
- Database capabilities
- clearinghouse of great lakes data of all kinds of omics data
- gather data on all animals
- visualize data in useful ways
- Link lab and field data
- integrate omics data and environmental data
- How to make this available to all
- students, practiioners, modelers, biologists, etc.
Conducting a Needs Assessment for the Great Lakes Peter C Esselman
* Native and non-native species use lake floor
* Most species have strong substrate type associations
* Very low density mapping of great lakes
* Solution
* [May want to try to get our lakefront designated as important mapping site
* Kendall et al 2018. Geosciences article based on geosicnece survey
* grid-based surveys from stakholders
* surveyed 24 individulas from 16 organizations
The Property Value Benefits of Remediating the Ashtabula River Area of Concern George Gardner
* 2.5 miles of river in AOC
* 2006, 2008, and 2013 cleanups spent 86 million
- Questions:
* How are values influenced?
* How do these match expectation?
- Data: