This repository is a alpha version of the new predict.sarlm() function, which will be included into spdep in R-forge later.

The goal is to extend prediction in spatial econometrics for in-sample and out-of-sample spatial units by implementing predictors proposed by papers of the References section below. For more detail, see this wiki page.

Run library(spdep) ; source("sarlm.R") ; source("nb2mat.R").

Examples are given in the sarlm.examples.R file.

The new documentation is provided in predict.sarlm.Rd.

1. Thomas-Agnan C, Laurent T, Goulard M (2014). "About predictions in spatial autoregressive models: Optimal and almost optimal strategies", TSE Working Paper, n. 13-452, December 18, 2013, revised September 2014.
2. Bivand RS (2002). "Spatial Econometrics Functions in R: Classes and Methods." Journal of Geographical Systems, 4, 405-421.
3. spatial-pred-r project
4. Kato, T. (2012). "Prediction in the lognormal regression model with spatial error dependence". Journal of Housing Economics 21: 66-76.
5. Kelejian, H. H. and I. R. Prucha (2007). "The relative efficiencies of various predictors in spatial econometric models containing spatial lags". Regional Science and Urban Economics 37: 363-374.
6. Kato (2008)
7. Griffith (2010) Spatial filtering and missing georeferenced data imputation: a comparison of the Getis and Griffith methods, in Perpectives on spatial data analysis, L. Anselin and S.J. Rey (eds), Springer-Verlag
8. LeSage and Pace (2004)

Notes on the previous references:

1. Definition of in-sample and out-of-sample prediction in the spatial case. Using its notations (eg. for the sub-spatial weight matrix). Proposed new predictors for the LAG model. Extended here to SDM model (and to SAC model, with warning ; to be confirmed).
2. Description of the previous version of predict.sarlm(). The pred.sarlm class may have to change.
3. Using its notations for new predictors. Using a different approach for predictors: for mixed models we include WX in the X matrix object. Our goal is to include this custom function sppred() into the predict.sarlm() framework.
4. Prediction with a log-transformed variable on SEM model. Anti-log transformation is biased: proposal of 4 other predictors to correct (partially or completely) this bias
5. 3 optimal predictors for the SAC model, based on 3 different information sets + 2 intuitive predictors (with one for the error model)
6. TODO: BPn predictor for SEM model?
7. EM procedure for missing value with Spatial Filtering
8. TODO: BP predictor in-sample for SEM model

• Bivand RS, Piras G (2015) "Comparing Implementations of Estimation Methods for Spatial Econometrics." Journal of Statistical Software, 63(18), 1-36
• Millo G, Piras G (2012). "splm: Spatial Panel Data Models in R." Journal of Statistical Software, 47(1), 1-38.
• Millo G (2014). "Maximum likelihood estimation of spatially and serially correlated panels with random effects". Computational Statistics & Data Analysis 71 (March), 914-933.
• Piras G (2010). "sphet: Spatial Models with Heteroskedastic Innovations in R." Journal of Statistical Software, 35(1), 1-21.

The following files are distributed under the GPL (>= 2) license (GPLv2 and GPLv3):

• sarlm.R
• sarlm.examples.R
• nb2mat.R
• predict.sarlm.Rd
• nb2mat.Rd

This project is partially based on the work of Jean-Sauveur AY, Raja CHAKIR and Julie LE GALLO