describe.rmd

---
title: " "
output: html_document
editor_options: 
  chunk_output_type: inline
---

```{r setup, include=FALSE}
carob_path = "../../../carob/"
# use package::function, do not use library()
#library(kableExtra)
#library(ggplot2)
#library(dplyr)
#library(viridis)

dataset <- "fertilizer"

datasets <- list.files(file.path(carob_path, "data", "clean", "fertilizer"), pattern = ".csv", full.names = TRUE)
datasets <- datasets[-grep("meta.csv", datasets)]

```

This is a test for data quality and exploratory analysis of carob data. The idea behind this is to provide some simple QA for each of the carob datasets in a summarized way in the [carob-data.org](http://carob-data.org/) page. 

With this, users can quickly have an overview of the data quality, and pick any strange behaviors in the data (e.g.: missing NA's), so that contributors can go and look closer at each of the datasets, view them and correct them (if deem necessary).

```{r plt, echo=FALSE, results = 'asis', fig.width = 8, fig.height = 3.25}

# Loop over each CSV file
for (file in csv_files) {
  
  # Print dataset_id
  cat(paste0("<p style='text-align: center; font-size:15pt;'>, ", gsub(".csv", "", basename(file)), "</p>\n\n"))

  # Read the clean dataset into a dataframe
  df <- utils::read.csv(file)
  
  # Get number of crops in dataset
  crops <- unique(df$crop)
  for (crop in seq_along(crops)) {
    # Subset by crop
    df2 <- df[df$crop == crops[crop],]
    # Create fertilizer classes/ranges
    df2$N_class <- cut(df2$N_fertilizer, breaks=c(-Inf, 40, 100, Inf), right = TRUE, labels = c("low", "mid", "high"))
    df2$P_class <- cut(df2$P_fertilizer, breaks=c(-Inf, 20, 50, Inf), right = TRUE, labels = c("low", "mid", "high"))
    df2$K_class <- cut(df2$K_fertilizer, breaks=c(-Inf, 20, 50, Inf), right = TRUE, labels = c("low", "mid", "high"))        
    # Plots
      n <- ggplot(df2, aes(x=N_class, y=yield)) +
        geom_violin(width = 0.5, scale = "count", na.rm = TRUE) +
        geom_point(position = position_jitter(width = 0.2), alpha = 0.1) +
        geom_boxplot(width=0.1, color="grey", outlier.alpha = 0) +
        scale_fill_viridis(discrete = TRUE) +
        ggtitle(paste0(crops[crop])) +
        labs(x = "Nitrogen level") +
        theme(plot.title = element_text(size = 12, face = "bold", hjust = 0.5))
      suppressWarnings(print(n))
      p <- ggplot(df2, aes(x=P_class, y=yield)) +
        geom_violin(width = 0.5, scale = "count", na.rm = TRUE) +
        geom_point(position = position_jitter(width = 0.2), alpha = 0.1) +
        geom_boxplot(width=0.1, color="grey", outlier.alpha = 0) +
        scale_fill_viridis(discrete = TRUE) +
        labs(x = "Phosphorous level") +
        theme(plot.title = element_text(size = 12, face = "bold", hjust = 0.5))
      suppressWarnings(print(p))
      k <- ggplot(df2, aes(x=K_class, y=yield)) +
        geom_violin(width = 0.5, scale = "count", na.rm = TRUE) +
        geom_point(position = position_jitter(width = 0.2), alpha = 0.1) +
        geom_boxplot(width=0.1, color="grey", outlier.alpha = 0) +
        scale_fill_viridis(discrete = TRUE) +
        labs(x = "Potassium level") +
        theme(plot.title = element_text(size = 12, face = "bold", hjust = 0.5))
      suppressWarnings(print(k))
  }

  # Add a page break after each histogram
  cat("\n\n")
  # cat("\\newpage\n")
}
```