Skip to content
NEMS helps you create & fit mathematical models to time series data
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
docs
nems
recordings
results
scripts
tests
unused
.gitignore
.travis.yml
LICENSE.txt
MANIFEST.in
README.md
fit_single.py
setup.py

README.md

NEMS

NEMS is the Neural Encoding Model System. It is helpful for fitting a mathematical model to time series data, plotting the model predictions, and comparing the predictive accuracy of multiple models. We use it to study computational models of how auditory stimuli are neurally encoded in mammalian brains, but it will probably work with your timeseries data as well.

Installation

git clone https://github.com/lbhb/NEMS

Then add the NEMS library via pip (where "NEMS" is the installation directory):

pip install -e NEMS

We have found that Python 3 distributions compiled with the Intel MKL libraries are about twice as fast as the default Python implementations that come installed on many linux machines. Please see our basic conda installation instructions if you would like to set up a python environment like the way that we do. NOTE: NEMS is designed to use Python 3. Backwards compatibility with Python 2 is untested and unsupported.

Alternatively (not recommended), you may install all the depdencies on your own, e.g.,

pip install requests numpy scipy matplotlib pandas

Then add NEMS to your python path. Eg, in Linux:

echo "export PYTHONPATH=\"\$PYTHONPATH:`pwd`/NEMS\"" >> ~/.bashrc
source ~/.bashrc

Your First Model Fit

You may test if everything is working by telling NEMS to download some sample auditory stimulus-response data, use a simple linear-nonlinear model (which should taking about 2 minutes to fit), and then save the results locally:

cd NEMS/scripts
ipython

In [1]: run demo_script.py

Open demo_script.py in an editor to work through each step of the fit.

Table of Contents

This documentation is a work in progress as of March 9, 2018.

  1. Quick Start
  2. Organizing your Data
  3. Organizing your Models
  4. Fitting your Models
  5. Detailed Guides
    • Architectural Diagram
    • Creating your own modules
    • Comparing your models with others
    • Sharing modules, models, and data with others
  6. Contributing to NEMS
  7. Other
You can’t perform that action at this time.