rlda-class: rlda_c
In CasAndreu/ldaRobust: LDA Robustness

Description Details Slots

Main class

methods: LDA wrapper, fit

dtm: document term matrix
lda_u: lda object(in topicmodels)
K: numeric or vector, if numeric, number of k to try, if vector, k's to try (will be overwrite with list of k's tried once fit has been run)
compute_parallel: bool, if true, will parallel when fitting lda models; otherwise, sequential for fitting
same_k_estimation: integer, number of initial states to try, if 0, do not try with diff state
similarity_measure: string, similarity measure (so far cosine or hellinger). Default: cosine
num_of_clusters: numeric or vector, number of clusters used when performing spectral clustering
seed_list: seeds tried (exists only when same_k_estimation is true)
beta_list: list of beta matrices in LDA objects of all K tried (ordered same as K)
gamma_list: list of gamma matrices in LDA objects of all K tried (ordered same as K)
model_topic_mat: percentage of documents dominated by the given topic
similarity_mat: maximum similarity (given choice of similarity functions) of a given topic compare to any topics in the original lda model (to give the probability of a user<e2><80><99>s topic shows up in a tried model<e2><80><99>s resulting topics)
sim_matrix_list: list of similarity matrices that gives us similarity between
init_states: # not used yet, probably gonna delete
key_features: top 10 features of a given topic in each model tried
topic_dom_perc_list: percentage of documents dominated by the given topic out of documents originally dominated by similar topic in the original model
dominant_topic_cluster_list: clusters correponding to dominant topics of each document in each model
cluster_center_key_words_list: (data_frame?) top 10 keywords for each center found by the cluster algorithn (so far only support spectral clustering)
perc_document_belong_cluster_list: percentage of documents belong to a given cluster in a given model
topic_cluster_assignment: cluster number a given topic belongs to d
top_stability_mat: a matrix indicating whether each cluster is present in each alternative model
doc_by_cluster_and_model: a matrix indicating the dominiant cluster of each document according to each topic model