| Title: | SVD-based Estimation for Exploratory Item Factor Analysis |
|---|---|
| Description: | Provides singular value decomposition based estimation algorithms for exploratory item factor analysis (IFA) based on multidimensional item response theory models. For more information, please refer to: Zhang, H., Chen, Y., & Li, X. (2020). A note on exploratory item factor analysis by singular value decomposition. Psychometrika, 1-15. |
| Authors: | Haoran Zhang [aut, cre], Yunxiao Chen [aut], Xiaoou Li [aut] |
| Maintainer: | Haoran Zhang <[email protected]> |
| License: | GPL-3 |
| Version: | 1.0 |
| Built: | 2024-11-01 03:37:50 UTC |
| Source: | https://github.com/hrzhang16/mirtsvd |
Item Factor Analysis by Singular Value Decomposition
mirtsvd(data, K, link = "logit", epsilon = 1e-04, rotation_fn = NULL, ...)mirtsvd(data, K, link = "logit", epsilon = 1e-04, rotation_fn = NULL, ...)
data |
the data matrix. Entries are either binary or categorical.
Missing entries should be |
K |
the number of factors. |
link |
the link fucntion. Possible choices are "logit" and "probit". |
epsilon |
the truncation parameter. Default value is 1e-4. |
rotation_fn |
rotation applied to the estimated loading matrix.
See |
... |
optional arguments passed to rotation_fn. |
The function returns a list with the following components:
The estimated loading matrix.
The rotation method.
The data type.
The number of categories in data.
Zhang, H., Chen, Y., & Li, X. (2020). A note on exploratory item factor analysis by singular value decomposition. Psychometrika, 1-15.
require(mirtjml) require(GPArotation) # load a simulated dataset attach(data_sim) data <- data_sim$response K <- data_sim$K res <- mirtsvd(data, K, rotation_fn = Varimax)require(mirtjml) require(GPArotation) # load a simulated dataset attach(data_sim) data <- data_sim$response K <- data_sim$K res <- mirtsvd(data, K, rotation_fn = Varimax)
Scree plot for singular values.
screeplot_svd(data, link = "logit", epsilon = 1e-04, K_max = 10)screeplot_svd(data, link = "logit", epsilon = 1e-04, K_max = 10)
data |
the data matrix. Entries are either binary or categorical.
Missing entries should be |
link |
the link fucntion. Possible choices are "logit" and "probit". |
epsilon |
the truncation parameter. Default value is 1e-4. |
K_max |
The maximum number of factors contained in data. Default value is 10. |
Zhang, H., Chen, Y., & Li, X. (2020). A note on exploratory item factor analysis by singular value decomposition. Psychometrika, 1-15.
require(mirtjml) # load a simulated dataset attach(data_sim) data <- data_sim$response screeplot_svd(data, K_max = 10)require(mirtjml) # load a simulated dataset attach(data_sim) data <- data_sim$response screeplot_svd(data, K_max = 10)