README.md

In Zanidean/autumn: Automated Many-Models

Automated Many Models in R

autumn is designed to streamline the many-models approach for exploratory data analysis and hypothesis testing.

• auto_mm() is a function to apply a model to many different subsets of data.

• model_maker() is a function factory to create a model function that can be applied with purrr.

• extract_model_metric() is a function for getting model metrics using the broom package.

Installation

devtools::install_github("zanidean/autumn")

Example

library(tidyverse)
library(autumn)

This will create a nested dataframe that has every subset of data, along with it’s models and corresponding model statistics.

df = diamonds %>% 
  auto_mm(model = model_maker(lm, price ~ x + y + z), # defines model to test
          split = c("cut", "color", "clarity"), # defines what to cut the data by
          permutations = 2) %>% # defined how many permutations of cuts to make
  extract_model_metric("p.value") %>% 
  extract_model_metric("r.squared")

This will create a plot of the many-models along a gradient corresponding to the quality statistic you define.

df %>% plot_mm(r.squared)

Zanidean/autumn documentation built on Oct. 31, 2019, 1:48 a.m.