Pipe Output of More Than One Variable Using tidyr::map or dplyr

In this article, we will explore how to create a list of 2X3X3 correlation matrices using the tidyr and dplyr packages in R. We will also discuss how to avoid redundancy in our code.

Introduction

The problem statement involves creating six correlation matrices that can be used to analyze the evolution of correlation between two variables, $spent and $quantity sold, over a period of three years. The data is stored in a data frame called DFile_Gather.

We will use the tidyr package to perform this task.

Background

The tidyr package is an extension to the tidyverse suite of packages, which includes the popular dplyr, ggplot2, and readr packages. The main purpose of tidyr is to provide tools for spreading data from long format to wide format and vice versa.

The map function in tidyr allows us to apply a function to each element of an input list, and the purrr package (a companion package to dplyr) provides a set of functions that can be used to map over lists.

In this article, we will use the map function from tidyr and purrr to create a list of 2X3X3 correlation matrices.

Solution

The problem statement mentions two variables: $spent and $quantity sold. These variables represent the values that we want to correlate with each other over time. The DFile_Gather data frame contains these values along with some additional information, such as calendar year and product type.

First, we will transform the DFile_Gather data frame by converting the Product_Type column to a factor using the transform function:

# Transform DFile_Gather into DG
DG <- transform(DFile_Gather, Product_Type = factor(Product_Type))

Next, we will split the transformed data frame into three lists: one for each calendar year. We can do this by splitting the data frame on the calendar year column using the split function:

# Split DG into s
s <- split(DG, DG$Calendar.Year)

We also need to define two variables that will be used as indices in our correlation calculations: $By and $Values. These variables represent the columns of interest in our data frame.

# Define By and Values
By <- c("Order.ID", "Product_Type")
Values <- c("Mexican_Pesos", "Quantity")

Now we can use a double Map function to calculate the correlation between each pair of variables. We will apply the cor function from the base R library to perform this calculation.

# Map over Values and s, and calculate correlations using cor
res <- Map(function(v) Map(function(s) cor(tapply(s[, v], s[By], c)), s), Values)

Finally, we will print out our results:

# Print the correlation matrix
print(res)

Output

The cor function returns a matrix of correlations between each pair of variables. The size of this matrix is 3x3 since there are three calendar years.

To calculate the correlation between two specific variables, we need to access the corresponding element in our result matrix.

# Accessing the correlation matrix for Quantity and Mexican_Pesos over all years
cor_matrix <- as.data.frame(res[[2]])
rownames(cor_matrix) <- c("Order.ID", "Product_Type")

We can also use this code to get the 3x3 correlation matrices for each of the two variables.

# Accessing the 3x3 correlation matrix for Mexican_Pesos over all years
cor_matrix_Mexican_Pesos <- as.data.frame(res[[1]])
rownames(cor_matrix_Mexican_Pesos) <- c("Order.ID", "Product_Type")

Conclusion

In this article, we have used the tidyr package in R to create a list of 2X3X3 correlation matrices between two variables. We avoided redundancy by using a double Map function and applying the cor function from base R.

We can use these correlation matrices to analyze the evolution of correlation over time for each pair of variables. This approach can be useful in various fields, such as finance or economics, where understanding the relationships between different variables is crucial.

Last modified on 2024-05-03