Recursive Quoted Language Expansion (oshka)
Expands quoted language by recursively replacing any symbol that points to quoted language with the language it points to. The recursive process continues until only symbols that point to non-language objects remain. The resulting quoted language can then be evaluated normally. This differs from the traditional ‘quote’/’eval’ pattern because it resolves intermediate language objects that would interfere with evaluation.

Automatic Short Form Creation (ShortForm)
Performs automatic creation of short forms of scales with an ant colony optimization algorithm. As implemented in the package, the algorithm randomly selects items to build a model of a specified length, then updates the probability of item selection according to the fit of the best model within each set of searches. The algorithm continues until the same items are selected by multiple ants a given number of times in a row. See Leite, Huang, & Marcoulides (2008) <doi:10.1080/00273170802285743> for an applied example.

Artificial Neural Networks for Anomaly Detection (ANN2)
Training of general classification and regression neural networks using gradient descent. Special features include a function for training autoencoders as well as an implementation of replicator neural networks, for details see Hawkins et al. (2012) <doi:10.1007/3-540-46145-0_17>. Multiple activation and cost functions (including Huber and pseudo-Huber) are included, as well as L1 and L2 regularization, momentum, early stopping and the possibility to specify a learning rate schedule. The package contains a vectorized gradient descent implementation which facilitates faster training through batch learning.

Greedy Stochastic Block Transition Models (GreedySBTM)
Performs clustering on the nodes of an undirected binary dynamic network, by maximising the exact integrated complete likelihood. The greedy algorithm used is described in Rastelli, R. (2017) ‘Exact integrated completed likelihood maximisation in a stochastic block transition model for dynamic networks’ <arXiv:1710.03551>.

Spatial Lag Model Trees (lagsarlmtree)
Model-based linear model trees adjusting for spatial correlation using a simultaneous autoregressive spatial lag.

Allocating Seats Methods and Party System Scores (electoral)
Highest averages & largest remainders allocating seats methods and several party system scores. Implemented highest averages allocating seats methods are D’Hondt, Webster, Danish, Imperiali, Hill-Huntington, Dean, Modified Sainte-Lague, equal proportions and Adams. Implemented largest remainders allocating seats methods are Hare, Droop, Hangenbach-Bischoff, Imperial, modified Imperial and quotas & remainders. The main advantage of this package is that ties are always reported and not incorrectly allocated. Party system scores provided are competitiveness, concentration, effective number of parties, party nationalization score, party system nationalization score and volatility. References. Gallagher (1991) <doi:10.1016/0261-3794(91)90004-C>. Norris (2004, ISBN:0-521-82977-1). Consejo Nacional Electoral del Ecuador (2014)<http://…/CAPITULO%206%20web.pdf>. Laakso & Taagepera (1979) <http://…/001041407901200101>. Jones & Mainwaring (2003) <https://…/304_0.pdf>. Pedersen (1979) <http://…/Pedersen.htm>.

Advertisements