A single code to create a grouped bar chart using ggplot

[sammigachuhi]

Hello,

I am still learning R. I was working on a census data in wide format, name as houseassets2. See the data in wide format below.

Id	County...Sub.County	Conventional.Households	Stand...alone.Radio	Desk.Top.Computer..Laptop..Tablet	Functional.Television	Analogue.Television	Internet	Bicycle	Motor.Cycle	Refrigerator	Car	Truck..Lorry..Bus..Three.Wheeler.truck	Tuk.Tuk
105	IKUTHA	16,579	54.6	1.9	10.7	1.9	6.4	30.5	13.8	1	1.9	0.6	0.6
106	KATULANI	12,166	56.5	3.2	22.4	3.4	7	25.8	12.5	3	4.5	0.7	0.4
107	KISASI	10,860	49	2.3	15	2.7	4.4	27.2	10.3	1.2	2.8	0.6	0.5
108	KITUI CENTRAL	29,026	55.7	8.7	37.2	5.1	15.5	13.3	12.1	8.6	6.3	1	0.7
109	KITUI WEST	17,465	60.3	3.8	27.3	4.5	7.3	28.7	15.5	2.8	5.4	0.7	0.7
110	KYUSO	15,969	49.4	2.4	13.2	1.9	6	21.8	14.7	1.1	2.7	0.8	0.4
111	LOWER YATTA	15,021	60.6	2.6	16.1	2.3	4.6	26.8	14	1.4	3.5	1	0.9
112	MATINYANI	11,497	60.7	3.4	25.7	3.1	4.7	23.3	14.9	2.8	3.9	0.6	0.8
113	MIGWANI	18,843	65.2	3.6	25.9	4.1	7.2	36.2	16.2	2.3	5.2	0.9	0.4
114	MUMONI	6,451	47.9	2.1	12.2	2.2	3.4	11.5	12.6	0.8	2.6	0.6	0.4
115	MUTITU	11,490	55.4	1.5	9.3	1.2	3.3	13.3	10.9	1	2.1	0.5	0.7
116	MUTITU NORTH	4,929	57.2	2.5	13.1	3	3.4	26.2	15.1	2	2.6	0.4	0.7
117	MUTOMO	23,034	52.1	2.5	10.4	1.9	5.5	24.3	13	1	2.1	0.6	0.5
118	MWINGI CENTRAL	26,745	54.1	4.5	24.7	3.7	8	17.6	13.2	3.7	4.6	0.8	0.5
119	MWINGI EAST	18,725	50.7	1.8	11	1.5	4.6	16.6	11.7	1.2	2.4	0.7	0.5
120	NZAMBANI	11,276	55.4	4	22	3.4	5.7	16.1	11	3.9	4.2	0.7	0.5
121	THAGICU	3,289	54.8	1.2	8.2	1.3	3.6	9.1	13.5	0.3	1.6	0.3	0.2
122	TSEIKURU	8,449	42.8	1.3	9.1	1.5	2.5	17	11.2	1	1.9	0.4	0.4

I used the melt function to convert the above table to long format in R studio. It is very large to past it here but it's accessible here
However, I run into problems trying to use a specific formula to select the values I wanted for my study from the variable column of the melted table. That is, the house assets refrigerators and internet. More like variables = c (Refrigerators, Internet) and other attempts was not working using ggplot.

##example
ggplot(assets.melt, aes(County...Sub.County, value, fill = assets.melt$variable[variable == 'Refrigerator', 'Internet'])) + geom_bar(stat = 'identity', position = 'dodge2') + scale_fill_manual(values = c('turquoise', 'rosybrown')) + theme(axis.text.x.bottom = element_text(size = 9, angle = 90, colour = 'blue'), legend.position = 'bottom' ) + xlab('Sub counties of Kitui')

I used the following steps to create the desired grouped bar plot:

1. select the specific columns you want to display from the wide table (in this case only three - Sub county, refrigerator, internet). ie.
   houseassets.kitui5 <- houseassets2 [(c(1, 7, 10))]
2. Used the melt function to transform the resulting table with just three columns above to a long format dataset. The melted table was assigned the name assets.melt2
3. Used ggplot to create a grouped bar chart with just the two variables displayed on the x axis for each County...sub county. Here is the code used after this long process:

#three step process worked but after a long process
ggplot(assets.melt2, aes(County...Sub.County, value, fill = variable)) + geom_bar(stat = 'identity', position = position_dodge2(width = 0.5)) + scale_fill_manual(values = c('turquoise', 'rosybrown')) + theme(axis.text.x.bottom = element_text(size = 9, angle = 90, colour = 'blue'), legend.position = 'bottom' ) + xlab('Sub-counties in Kitui') + ylab ('Percentage of households') + labs (title = 'Percentage of households with selected assets in Kitui County') + guides(fill = guide_legend(title = 'Type of assets'))

The resulting bar chart after this 3 step process is shown below.

So to sum it up, I am just wondering if there is a way I could write in a single code, just after melting the entire wide table, a code that choses the specific values from a column to display in the bar chart like the one above without going through the detailed 3 step process.
ggplot(assets.melt2, aes(County...Sub.County, value, fill = <what do I fill here or some other way)) + geom_bar(stat = 'identity', position = position_dodge2(width = 0.5))

I prefer using ggplot than the plot() defaults. Thanks

[jonspring]

I'm not sure I'm understanding your question, but it sounds like dplyr::filter might help here. Then you can use one long data frame for all cases and filter it before it goes into ggplot2 rather than generating custom tables for each combination of categories.

library(dplyr)
filter_items <- c("Refrigerator", "Internet")
assets.melt %>%
   filter(variable %in% filter_items) %>%
   ggplot(aes(County...Sub.County, value, fill = variable)) +
  geom_col(position = 'dodge2') +
  #etc...