Bayesian Item Response Theory Estimation.
Project description
GIRTH MCMC
Item Response Theory using Markov Chain Monte Carlo / Variational Inference
Dependencies
We recommend using Anaconda. Individual packages can be installed through pip otherwise.
- Python ≥ 3.8
- Numpy
- Scipy
- Girth
- PyMC3
Installation
Via pip
pip install girth_mcmc --upgrade
From Source
pip install . -t $PYTHONPATH --upgrade
Supports
Unidimensional
- Rasch Model
- 1PL Model
- 2PL Model
- 3PL Model
- Graded Response Model
- Partial Credit Model
Multi-dimensional
- 2PL Model
- Graded Response Model
- Partial Credit Model
Usage
Subject to change but for now:
import numpy as np
from girth.synthetic import create_synthetic_irt_dichotomous
from girth_mcmc import GirthMCMC
discrimination = 0.89 * np.sqrt(-2 * np.log(np.random.rand(10)))
difficulty = np.random.randn(10)
theta = np.random.randn(100)
syn_data = create_synthetic_irt_dichotomous(difficulty, discrimination,
theta)
girth_model = GirthMCMC(model='2PL',
options={'n_processors': 4})
results = girth_model(syn_data)
print(results)
for the graded response model, pass in the number of categories
import numpy as np
from girth.synthetic import create_synthetic_irt_polytomous
from girth_mcmc import GirthMCMC
n_categories = 3
difficulty = np.random.randn(10, n_categories-1)
difficulty = np.sort(difficulty, 1)
discrimination = 0.96 * np.sqrt(-2 * np.log(np.random.rand(10)))
theta = np.random.randn(150)
syn_data = create_synthetic_irt_polytomous(difficulty, discrimination,
theta, model='grm')
girth_model = GirthMCMC(model='GRM', model_args=(n_categories,),
options={'n_processors': 4})
results = girth_model(syn_data)
print(results)
Is some data missing? Tag it with a convenience function and run it like normal
import numpy as np
from girth.synthetic import create_synthetic_irt_dichotomous
from girth_mcmc import GirthMCMC
from girth_mcmc.utils import tag_missing_data_mcmc
discrimination = 0.89 * np.sqrt(-2 * np.log(np.random.rand(10)))
difficulty = np.random.randn(10)
theta = np.random.randn(100)
syn_data = create_synthetic_irt_dichotomous(difficulty, discrimination,
theta)
mask = np.random.rand(*syn_data.shape) < .1
syn_data[mask] = -9999
syn_data_missing = tag_missing_data_mcmc(syn_data, [0, 1])
girth_model = GirthMCMC(model='2PL',
options={'n_processors': 4})
results = girth_model(syn_data_missing)
print(results)
Don't like waiting? me either. Run Variational Inference for faster but less accurate estimation.
import numpy as np
from girth.synthetic import create_synthetic_irt_polytomous
from girth_mcmc import GirthMCMC
n_categories = 3
difficulty = np.random.randn(10, n_categories-1)
difficulty = np.sort(difficulty, 1)
discrimination = 1.76 * np.sqrt(-2 * np.log(np.random.rand(10)))
theta = np.random.randn(150)
syn_data = create_synthetic_irt_polytomous(difficulty, discrimination,
theta, model='grm')
girth_model = GirthMCMC(model='GRM', model_args=(n_categories,),
options={'variational_inference': True,
'variational_samples': 10000,
'n_samples': 10000})
results_variational = girth_model(syn_data, progressbar=False)
print(results_variational)
Unittests
pytest with coverage.py module
pytest --cov=girth_mcmc --cov-report term
Contact
Ryan Sanchez
ryan.sanchez@gofactr.com
Other Estimation Packages
If you are looking for Marginal Maximum Likelihood estimation routines, check out GIRTH, a graphical interface is also at GoFactr
License
MIT License
Copyright (c) 2021 Ryan C. Sanchez
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file girth_mcmc-0.6.0.tar.gz
.
File metadata
- Download URL: girth_mcmc-0.6.0.tar.gz
- Upload date:
- Size: 15.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.6.0 importlib_metadata/4.8.2 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.9.8
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | fa9df9b47627e33ec3c962e3176dc13178f86a28072d048ce0371eef925a6457 |
|
MD5 | dbae300f934f3472cbbfffc9128658c7 |
|
BLAKE2b-256 | 4b3304b00ea9dedbefc24124555f8483102eebd5adaed49fbedfa483a319b97b |
File details
Details for the file girth_mcmc-0.6.0-py3-none-any.whl
.
File metadata
- Download URL: girth_mcmc-0.6.0-py3-none-any.whl
- Upload date:
- Size: 21.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.6.0 importlib_metadata/4.8.2 pkginfo/1.7.1 requests/2.26.0 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.9.8
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | fe9a70435ef75bd7f71c43ab280e7a1417c666d135a3f94201c78eca121fe951 |
|
MD5 | 142a9f2c6554d4529b997c3708b07cb2 |
|
BLAKE2b-256 | 053bca2fb8af758402795ec55aeeb78a36f939d4c870349a101d1e12fb50ba58 |