The goal of ln.knn.regression is to assist in creating a regression analysis project from start to finish. It includes R functions that help to explore, summarize and visualize the data. It also includes functions that directly build and visualize linear and k-nn models. Additionally, it contains functions that can test the model’s error and accuracy.

ln.knn.regression is similar to tidymodels which is a collection of packages for modeling and machine learning using tidyverse principles. ln.knn.regression imports many of the same packages that tidymodels and tidyverse use, such as dplyr, ggplot2, parsnip, rsample and more. Like tidymodels, ln.knn.regression contains functions for building models and performing data analysis. However, ln.knn.regression is a smaller and more simple package that is used specifically for regression modeling and analysis.

You can install the development version of ln.knn.regression from GitHub with:

# install.packages("devtools")
devtools::install_github("DSCI-310/dsci-310-group-09-pkg")

• bar_graph()
• hist_plot()
• knn_model()
• linearmodel()
• model_rmspe()
• scatter_plot()
• summarize_column()

Here is a basic example which shows you how to explore the data and build a linear regression model.

library(ln.knn.regression)

summarize_column(mtcars, mtcars$mpg)
#>       mean  med       sd
#> 1 20.09062 19.2 6.026948

hist_plot(mtcars, mpg, "MPG", "Count", "Histogram for MPG of Cars", 10)
#> `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

mtcars_model <- linearmodel(recipes::recipe(mpg ~ hp, mtcars_training), mtcars_training)
mtcars_model
#> ══ Workflow [trained] ══════════════════════════════════════════════════════════
#> Preprocessor: Recipe
#> Model: linear_reg()
#> 
#> ── Preprocessor ────────────────────────────────────────────────────────────────
#> 0 Recipe Steps
#> 
#> ── Model ───────────────────────────────────────────────────────────────────────
#> 
#> Call:
#> stats::lm(formula = ..y ~ ., data = data)
#> 
#> Coefficients:
#> (Intercept)           hp  
#>    28.36292     -0.05863

scatter_plot(mtcars_training, 
             hp, 
             mpg, 
             "Horse Power (HP)", 
             "Miles Per Gallon (mpg)",
             "Best Fit Line for MPG vs HP", 
             15,
             "lm") 
#> `geom_smooth()` using formula = 'y ~ x'

mtcars_rmspe <- model_rmspe(mtcars_model, mtcars_testing, "mpg")
mtcars_rmspe
#> [1] 4.99694

This tells us that the model has a prediction error of around 5 miles per gallon when tested on data it has not seen before.

For a detailed tutorial on how to use each of the functions in ln.knn.regression, refer to our Vignette.

To view an example of when this package was used for an analysis, please view this repository.

We appreciate contributions to the ln.knn.regression package. Please refer to our Contributing document and Code of Conduct for more information.

The software provided in this project is offered under the MIT open source license. Refer to the license file for more information.