lm.md

In 2DegreesInvesting/ds-targets: What the Package Does (One Line, Title Case)

A linear model suggests ozone levels are stable

Setup

library(targets)

Introduction

The bulk of the computation should have already happened upstream, and the most of the code chunks in the report itself should be terse calls to tar_read() and tar_load().

–https://books.ropensci.org/targets/files.html#literate-programming

This report summarizes our results, mostly as terse calls to tar_read(). For implementation details, see the underlying targets pipeline:

writeLines(readLines("_targets.R"))
#> library(targets)
#> library(tarchetypes)
#> 
#> tar_option_set(imports = "ds.targets", packages = c("ds.targets"))
#> 
#> list(
#>   # External
#>   tar_target(path, raw_path(), format = "file"),
#> 
#>   # Internal: _targets/objects/
#>   tar_target(raw, command = read(path)),
#>   tar_target(data, clean(raw)),
#>   tar_target(lm_fit, fit_lm(data)),
#>   tar_target(lm_plot, plot_lm(data)),
#> 
#>   # External
#>   tar_target(lm_figure, save_plot(path = "output/plot.png", plot = lm_plot)),
#>   tarchetypes::tar_render(lm, "lm.Rmd", output_format = "md_document"),
#>   tarchetypes::tar_render(plot, "plot.Rmd", output_format = "md_document")
#> )

Methods

To make this project easy to reproduce and share we built the R package ds.targets. You can install it from GitHub with:

# install.packages("devtools")
devtools::install_github("2DegreesInvesting/ds.targets")

To clean the raw data we removed observations where Ozone values were missing.

# Source code
ds.targets::clean
#> function(raw) {
#>   dplyr::mutate(
#>     raw,
#>     Ozone = tidyr::replace_na(.data$Ozone, mean(.data$Ozone, na.rm = TRUE))
#>   )
#> }
#> <bytecode: 0x4d0b088>
#> <environment: namespace:ds.targets>

# Before
raw_data <- tar_read(raw)
any(is.na(raw_data$Ozone))
#> [1] TRUE

# After
clean_data <- tar_read(data)
any(is.na(clean_data$Ozone))
#> [1] FALSE

To explored changes in Ozone through time we use a linear model.

model <- tar_read(lm_fit)

Results

Here are our results:

summary(model)
#> 
#> Call:
#> stats::lm(formula = Ozone ~ Day, data = data)
#> 
#> Residuals:
#>     Min      1Q  Median      3Q     Max 
#> -40.938 -21.269  -0.434   3.878 126.209 
#> 
#> Coefficients:
#>             Estimate Std. Error t value Pr(>|t|)    
#> (Intercept) 42.71020    4.76908   8.956 1.17e-15 ***
#> Day         -0.03676    0.26340  -0.140    0.889    
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> 
#> Residual standard error: 28.79 on 151 degrees of freedom
#> Multiple R-squared:  0.0001289,  Adjusted R-squared:  -0.006493 
#> F-statistic: 0.01947 on 1 and 151 DF,  p-value: 0.8892

Conclusion

We see no evidence that Ozone changes through time.

2DegreesInvesting/ds-targets documentation built on March 19, 2021, 9:45 a.m.