NEWS.md

In newsmap: Semi-Supervised Model for Geographical Document Classification

Changens in v0.8.3

• Fix for changes in the Matrix package v1.5

Changens in v0.8.2

• Add min_conf to predict()
• Add experimental argument entropy to textmodel_newsmap()

Changens in v0.8.1

• Update the seed words for UA (#64)

Changens in v0.8.0

• Add Portuguese dictionary
• Add "Saigon" to VE in selected languages (#47)
• Add label and drop_label to textmodel_newsmap()
• Add rescale and min_n to predict()
• Add print() methods for fitted models

Changens in v0.7.4

• Change predicted values to factor with all labels in training data

Changes in v0.7.3

• Update tests for quanteda v3.0

Changes in v0.7.2

• Make predict() significantly faster

Changes in v0.7.1

• Improve efficiency of textmodel_newsmap()
• Add compatibility with newer textstat_entropy()

Changes in v0.7.0

• Add Hebrew and Arabic seed dictionaries

Changes in v0.6.9

• Clean up Italian, German and Spanish seed dictionaries

Changes in v0.6.8

• Add Italian seed dictionary

Changes in v0.6.7

Dictionary updates

• Correct Japanese seed words for DE, MG and EC

Changes in v0.6.4

Bug fixes

• Return NA for documents that do not have known features
• Drop document variables to avoid slowdown and warnings

New data

• Add Chinese (simplified and traditional) seed dictionaries
• Add French seed dictionary

New function

• Add a function to compute average feature entropy (AFE)

Changes in v0.6.0

Bug fixes

• Fix error in textmodel_newsmap() when smooth is < 1.0
• Fitted models no longer include classes that did not occur in training set

Changes in v0.4.6

New functions

• Add coef() and coefficients() methods

Dictionary updates

• Add Russian language seed dictionary
• Correct Cook Islands' country code from CC to CK, and remove it from POL
• De facto capital cities of TZ, ZA, NG and BO are added to all dictionaries
• CC is added to DE, JA and EN dictionaries
• Dictionary entries are sorted in alphabetical order of country code

newsmap documentation built on Oct. 7, 2023, 5:07 p.m.