I am new to shiny/flexdashboard and so far have been able to render plots and filter dataframe by using values from selectInput with help of req(input$user_input_value) .
ISSUE: To run kmeans I am taking user input for number of clusters which I am not able to code it in reactive format and getting error: object of type closure is not subsettable.
```{r setup, include=FALSE}
library(flexdashboard)
library(shiny)
library(tidyverse)
library(tidytext)
library(scales)
library(glue)
library(widyr)
library(factoextra)
```
df
1 2 3 4
Angola -0.08260540 0.034325891 -0.02013353 -0.014063951
Armenia -0.06613693 -0.044308626 -0.13230387 -0.024534033
Azerbaijan -0.07562365 -0.003670707 0.05886792 -0.219660410
Bahrain -0.08275891 0.035843793 -0.02280102 -0.008044934
Bangladesh -0.08306371 0.032998297 -0.02634819 -0.017627316
Bosnia & Herzegovina -0.06303898 -0.050781511 -0.15183954 0.016794674
(Note: I have placed the csv file in github & mentioned its link below. For kmeans the character column should be used as rownames which represents country here.)
UPDATED df CREATION STEP
svd_dimen_all_wide <- read.csv(url("https://raw.githubusercontent.com/johnsnow09/covid19-df_stack-code/main/svd_dimen_all_wide.csv"))
svd_dimen_all_wide <- as.data.frame(svd_dimen_all_wide)
rownames(svd_dimen_all_wide) <- svd_dimen_all_wide$X
svd_dimen_all_wide <- svd_dimen_all_wide[,2:ncol(svd_dimen_all_wide)]
flexdashboard
---
title: "UN Country Votes"
output:
flexdashboard::flex_dashboard:
orientation: columns
vertical_layout: fill
theme: space
runtime: shiny
---
```{r setup, include=FALSE}
library(flexdashboard)
library(shiny)
library(tidyverse)
library(tidytext)
library(scales)
library(glue)
library(widyr)
library(factoextra)
Page NAme
=====================================
Inputs {.sidebar}
-----------------------------------------------------------------------
```{r}
selectInput("number_of_clusters", label = h3("Number of Clusters"),
choices = c(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15) ,
selected = 6)
```
Column {data-width=1000}
-----------------------------------------------------------------------
```{r include=FALSE}
set.seed(123)
km.res <- reactive({
# req(input$number_of_clusters)
kmeans(svd_dimen_all_wide, as.numeric(input$number_of_clusters), nstart = 25)
})
df_with_cluster <- cbind(svd_dimen_all_wide, cluster = km.res$cluster)
df_with_cluster <- rownames_to_column(df_with_cluster, "country")
df_with_cluster <- df_with_cluster %>%
select(country, cluster, everything())
```
UPDATED ATTEMPT:
renderPrint({
df_with_cluster <- cbind(svd_dimen_all_wide, cluster = km.res()$cluster)
df_with_cluster <- rownames_to_column(df_with_cluster, "country")
df_with_cluster <- df_with_cluster %>%
select(country, cluster, everything())
head(df_with_cluster)
})
### Comparison of Countries on Yes% of Bi Words
```{r}
renderPlot({
world_data %>%
left_join((df_with_cluster %>%
mutate(country_code = countrycode(country, "country.name", "iso2c"))
),
by = c("country_code")) %>%
filter(!is.na(cluster)) %>%
ggplot(aes(x = long, y = lat, group = group,
fill = as.factor(cluster))) +
geom_polygon() +
theme_map() +
scale_fill_discrete() +
labs(fill = "cluster",
title = "World Clusters based on UN voting",
caption = "created by ViSa") +
theme(plot.title = element_text(face = "bold", size = 16))
})
```
The problem is in a reactive chunk. The reactive expression km.res uses an input number of clusters, runs a model, and saves the output. (and let's end the code chunk here).
Next, decide what do you want to do with the output?
Now Let's print the output of the model with renderPrint() the output can be accessed by calling the expression’s name followed by parenthesis, e.g., km.res()
Column {data-width=1000}
-----------------------------------------------------------------------
```{r include=FALSE}
km.res <- reactive({
req(input$number_of_clusters)
set.seed(123)
kmeans(svd_dimen_all_wide, as.numeric(input$number_of_clusters), nstart = 25)
})
```
###
```{r model}
renderPrint({
df_with_cluster <- cbind(svd_dimen_all_wide, cluster = km.res()$cluster)
head(df_with_cluster)
})
```
Here is my blog post very relevant to this problem https://towardsdatascience.com/build-an-interactive-machine-learning-model-with-shiny-and-flexdashboard-6d76f59a37f9?sk=922526470699966c3f47b24843404a15