FDR-Control in Multiscale Change-Point Segmentation (FDRSeg)
Estimate step functions via multiscale inference with controlled false discovery rate (FDR). For details see H. Li, A. Munk and H. Sieling (2016) <doi:10.1214/16-EJS1131>.

Call Google’s ‘Natural Language’ API, ‘Cloud Translation’ API and ‘Cloud Speech’ API (googleLanguageR)
Call ‘Google Cloud’ machine learning APIs for text and speech tasks. Call the ‘Cloud Translation’ API <https://…/> for detection and translation of text, the ‘Natural Language’ API <https://…/> to analyse text for sentiment, entities or syntax or the ‘Cloud Speech’ API <https://…/> to transcribe sound files to text.

Random Network Model Selection and Parameter Tuning (randnet)
Model selection and parameter tuning procedures for a class of random network models. The model selection can be done by a general cross-validation framework called ECV from Li et. al. (2016) <arXiv:1612.04717> . Several other model-based and task-specific methods are also included, such as NCV from Chen and Lei (2016) <arXiv:1411.1715>, likelihood ratio method from Wang and Bickel (2015) <arXiv:1502.02069>, spectral methods from Le and Levina (2015) <arXiv:1507.00827>. Many network analysis methods are also implemented, such as the regularized spectral clustering (Amini et. al. 2013 <doi:10.1214/13-AOS1138>) and its degree corrected version and graphon neighborhood smoothing (Zhang et. al. 2015 <arXiv:1509.08588>).

Alternative Bootstrap-Based t-Test Aiming to Reduce Type-I Error for Non-Negative, Zero-Inflated Data (rbtt)
Tu & Zhou (1999) <doi:10.1002/(SICI)1097-0258(19991030)18:20%3C2749::AID-SIM195%3E3.0.CO;2-C> showed that comparing the means of populations whose data-generating distributions are non-negative with excess zero observations is a problem of great importance in the analysis of medical cost data. In the same study, Tu & Zhou discuss that it can be difficult to control type-I error rates of general-purpose statistical tests for comparing the means of these particular data sets. This package allows users to perform a modified bootstrap-based t-test that aims to better control type-I error rates in these situations.

Local Interpretable Model-Agnostic Explanations (lime)
When building complex models, it is often difficult to explain why the model should be trusted. While global measures such as accuracy are useful, they cannot be used for explaining why a model made a specific prediction. ‘lime’ (a port of the ‘lime’ ‘Python’ package) is a method for explaining the outcome of black box models by fitting a local model around the point in question an perturbations of this point. The approach is described in more detail in the article by Ribeiro et al. (2016) <arXiv:1602.04938>.

Analysis of Data with Mixed Measurement Error and Misclassification in Covariates (augSIMEX)
Implementation of the augmented Simulation-Extrapolation (SIMEX) algorithm proposed by Yi et al. (2015) <doi:10.1080/01621459.2014.922777> for analyzing the data with mixed measurement error and misclassification. The main function provides a similar summary output as that of glm() function. Both parametric and empirical SIMEX are considered in the package.

Advertisements