Descriptive Statistics for Categorial Data

library("dplyr")
library("ggplot2")
library("readxl")
library("DT")

## importing data

transac_data <- read_xlsx(path = "Data/Supermarket_transactions.xlsx",
                          sheet = 2)

Frequencies

Counts for different combinations of categorical variables

## Contingency tables
attach(transac_data)
table(Gender, `Marital Status`)

##       Marital Status
## Gender    M    S
##      F 3602 3568
##      M 3264 3625

## Multidimensional tables based  on three or more categorical variables

table1 <- table(Gender, `Marital Status`,`State or Province`)
ftable(table1)

##                       State or Province   BC   CA   DF Guerrero Jalisco   OR Veracruz   WA Yucatan Zacatecas
## Gender Marital Status                                                                                       
## F      M                                 190  638  188       77      15  510      142 1166     200       476
##        S                                 183  686  175      107      30  607      125 1134     164       357
## M      M                                 197  692  210       94       5  514      108 1160     129       155
##        S                                 239  717  242      105      25  631       89 1107     161       309

Proportions

Distribution of the categories in a variable.

table2 <- table(Gender) 

datatable(data = as_tibble(prop.table(table2)) %>% 
                    rename(Proportion = n) %>%
                    mutate(Proportion = round(Proportion*100,3)),
          style = "Bootstrap")

# percentages for gender by marital status
table3 <- table(`Marital Status`, Gender)

datatable(data = as_tibble(prop.table(table3)) %>% 
                    rename(Proportion = n) %>%
                    mutate(Proportion = round(Proportion*100,3)),
          style = "Bootstrap")

# customer percentages across location by gender and marital status
# using table1 from previous code chunk
ftable(round(prop.table(table1), 3))

##                       State or Province    BC    CA    DF Guerrero Jalisco    OR Veracruz    WA Yucatan Zacatecas
## Gender Marital Status                                                                                            
## F      M                                0.014 0.045 0.013    0.005   0.001 0.036    0.010 0.083   0.014     0.034
##        S                                0.013 0.049 0.012    0.008   0.002 0.043    0.009 0.081   0.012     0.025
## M      M                                0.014 0.049 0.015    0.007   0.000 0.037    0.008 0.083   0.009     0.011
##        S                                0.017 0.051 0.017    0.007   0.002 0.045    0.006 0.079   0.011     0.022

Marginals

Marginals show the total counts or percentages across columns or rows in a contingency table:

margin.table(table3, 1) ## sum Across Marital status (Rows)

## Marital Status
##    M    S 
## 6866 7193

margin.table(table3, 2) ## Sum across Gender (Column)

## Gender
##    F    M 
## 7170 6889

## Proportions across MArital status (Row sum equal to one)
prop.table(table3, margin = 1)

##               Gender
## Marital Status         F         M
##              M 0.5246140 0.4753860
##              S 0.4960378 0.5039622

 ## Proportions across gender (Colsum equal to one)
prop.table(table3, margin = 2)

##               Gender
## Marital Status         F         M
##              M 0.5023710 0.4737988
##              S 0.4976290 0.5262012

Visualization

reorder_size <- function(x) {
        factor(x, levels = names(sort(table(x), decreasing = TRUE)))
}

ggplot(transac_data, aes(x = reorder_size(`State or Province`))) +
        geom_bar(fill = "#ed5107") +
        theme_classic() +
        theme(axis.text.x = element_text(angle = 45, hjust = 1),
              plot.title = element_text(hjust = 0.5)) +
        ggtitle("Bar Charts")

## Or for proportions

ggplot(transac_data, aes(x = reorder_size(`State or Province`))) +
        geom_bar(fill = "#02383c", aes( y = (..count..)/sum(..count..))) +
        theme_classic() +
        scale_y_continuous(labels = scales::percent, name = "Proportion") +
        theme(axis.text.x = element_text(angle = 45, hjust = 1),
              plot.title = element_text(hjust = 0.5)) +
        ggtitle("Bar Charts")

### Multiple Graph splitted by category

ggplot(transac_data, aes(x = reorder_size(`State or Province`))) +
        geom_bar(aes(y = (..count..)/sum(..count..)), fill = "#ed0cef") +
        xlab("State or Province") +
        scale_y_continuous(labels = scales::percent, name = "Proportion") +
        facet_grid(`Marital Status`~ Gender) +
        theme(axis.text.x = element_text(angle = 45, hjust = 1)) + 
        theme_classic()