R：图上的重叠点

Question

I am using the R programming language. I am trying to learn how to overlay points on a graph and then visualize them.

Using the following code, I can generate some time series data, aggregate them by month, taking the average/min/max, and plot the following graph:

library(xts)
library(ggplot2)
library(dplyr)
library(plotly)
library(lubridate)

set.seed(123)

#time series 1
date_decision_made = seq(as.Date("2014/1/1"), as.Date("2016/1/1"),by="day")

property_damages_in_dollars <- rnorm(731,100,10)

final_data <- data.frame(date_decision_made, property_damages_in_dollars)


#####aggregate

final_data$year_month <- format(as.Date(final_data$date_decision_made), "%Y-%m")
final_data$year_month <- as.factor(final_data$year_month)


f = final_data %>% group_by (year_month) %>% summarise(max_value = max(property_damages_in_dollars), mean_value = mean(property_damages_in_dollars), min_value = min(property_damages_in_dollars))



####plot####

fig <- plot_ly(f, x = ~year_month, y = ~max_value, type = 'scatter', mode = 'lines',
        line = list(color = 'transparent'),
        showlegend = FALSE, name = 'max_value') 

fig <- fig %>% add_trace(y = ~min_value, type = 'scatter', mode = 'lines',
            fill = 'tonexty', fillcolor='rgba(0,100,80,0.2)', line = list(color = 'transparent'),
            showlegend = FALSE, name = 'min_value') 

fig <- fig %>% add_trace(x = ~year_month, y = ~mean_value, type = 'scatter', mode = 'lines',
            line = list(color='rgb(0,100,80)'),
            name = 'Average') 


fig <- fig %>% layout(title = "Average Property Damages",
         paper_bgcolor='rgb(255,255,255)', plot_bgcolor='rgb(229,229,229)',
         xaxis = list(title = "Months",
                      gridcolor = 'rgb(255,255,255)',
                      showgrid = TRUE,
                      showline = FALSE,
                      showticklabels = TRUE,
                      tickcolor = 'rgb(127,127,127)',
                      ticks = 'outside',
                      zeroline = FALSE),
         yaxis = list(title = "Dollars",
                      gridcolor = 'rgb(255,255,255)',
                      showgrid = TRUE,
                      showline = FALSE,
                      showticklabels = TRUE,
                      tickcolor = 'rgb(127,127,127)',
                      ticks = 'outside',
                      zeroline = FALSE))

fig

Now (on the same plot "fig"), for each month, I am trying to plot all the observations for that month in a vertical fashion. I am trying to create something like this:

With a bit of data manipulation, the following code can produce the graph below: plot( final_data$year_month, final_data$property_damages_in_dollars)

Can someone please show me how to extend this solution for a plotly diagram (ie enhance the "fig" object)?

Thanks

Answer 1

To have full flexibilty with regards to formatting your markers, you can use add_trace with subsets of your dataframe final_data using the following addition to your code:

date_split <- split(final_data, final_data$year_month)
for (i in 1:length(date_split)) {
  fig <- fig %>% add_trace(y=date_split[[i]]$property_damages_in_dollars,
                           x=date_split[[i]]$year_month,
                           mode='markers'
                           )
}

Result 1:

If you'd like only black markers you can add the following to add_trace() :

marker=list(color='rgba(0,0,0, 1)'

Result 2:

And if you'd like to adjust the transparency of your plots you can do so directly throug tha last argument in rgba() , for example:

marker=list(color='rgba(0,0,0, 0.2)')

Result 3:

Complete code:

library(xts)
library(ggplot2)
library(dplyr)
library(plotly)
library(lubridate)

set.seed(123)

#time series 1
date_decision_made = seq(as.Date("2014/1/1"), as.Date("2016/1/1"),by="day")

property_damages_in_dollars <- rnorm(731,100,10)

final_data <- data.frame(date_decision_made, property_damages_in_dollars)


#####aggregate

final_data$year_month <- format(as.Date(final_data$date_decision_made), "%Y-%m")
final_data$year_month <- as.factor(final_data$year_month)


f = final_data %>% group_by (year_month) %>% summarise(max_value = max(property_damages_in_dollars), mean_value = mean(property_damages_in_dollars), min_value = min(property_damages_in_dollars))



####plot####

fig <- plot_ly(f, x = ~year_month, y = ~max_value, type = 'scatter', mode = 'lines',
        line = list(color = 'transparent'),
        showlegend = FALSE, name = 'max_value') 

fig <- fig %>% add_trace(y = ~min_value, type = 'scatter', mode = 'lines',
            fill = 'tonexty', fillcolor='rgba(0,100,80,0.2)', line = list(color = 'transparent'),
            showlegend = FALSE, name = 'min_value') 

fig <- fig %>% add_trace(x = ~year_month, y = ~mean_value, type = 'scatter', mode = 'lines',
            line = list(color='rgb(0,100,80)'),
            name = 'Average') 


fig <- fig %>% layout(title = "Average Property Damages",
         paper_bgcolor='rgb(255,255,255)', plot_bgcolor='rgb(229,229,229)',
         xaxis = list(title = "Months",
                      gridcolor = 'rgb(255,255,255)',
                      showgrid = TRUE,
                      showline = FALSE,
                      showticklabels = TRUE,
                      tickcolor = 'rgb(127,127,127)',
                      ticks = 'outside',
                      zeroline = FALSE),
         yaxis = list(title = "Dollars",
                      gridcolor = 'rgb(255,255,255)',
                      showgrid = TRUE,
                      showline = FALSE,
                      showticklabels = TRUE,
                      tickcolor = 'rgb(127,127,127)',
                      ticks = 'outside',
                      zeroline = FALSE))

date_split <- split(final_data, final_data$year_month)
for (i in 1:length(date_split)) {
  fig <- fig %>% add_trace(y=date_split[[i]]$property_damages_in_dollars,
                           x=date_split[[i]]$year_month,
                           mode='markers',
                           marker=list(color='rgba(0,0,0, 0.2)')
                           #marker=list(color='rgba(0,0,0, 1)')
                           )
}
fig

Answer 2

At least I always find it simpler to go with ggplot and then send it to plotly by using the magical function ggplotly . Hopefully this helps you.

library(xts)
library(ggplot2)
library(dplyr)
library(plotly)
library(lubridate)

set.seed(123)

#time series 1
date_decision_made = seq(as.Date("2014/1/1"), as.Date("2016/1/1"),by="day")

property_damages_in_dollars <- rnorm(731,100,10)

final_data <- data.frame(date_decision_made, property_damages_in_dollars)


#####aggregate

dat <- final_data %>% 
  mutate(month = month(date_decision_made),
         year = year(date_decision_made),
         month_end = ceiling_date(date_decision_made, unit = "month")-1) %>% 
  group_by(month, year) %>% 
  mutate(mean_val = mean(property_damages_in_dollars,na.rm = TRUE),
         max_val = max(property_damages_in_dollars,na.rm = TRUE),
         min_val = min(property_damages_in_dollars,na.rm = TRUE))

p <- ggplot(data = dat) +
  geom_ribbon(aes(x = month_end, 
                  ymin = min_val,
                  ymax = max_val), alpha = 0.2)+
  geom_point(aes(x = month_end,
             y = property_damages_in_dollars), alpha = 0.3) +
  geom_line(aes(x = month_end,
                y = mean_val), size = 1.25) +
  labs(y = "Dollars",
       x = "Months")+
  theme_minimal()
  
ggplotly(p)

Answer 3

The following addition to your last line of code:

fig %>% add_trace(data = final_data, 
              y = ~property_damages_in_dollars, x = ~year_month, 
              name = "Property Damage in Dollars", mode = "markers", 
              marker = list(color = " rgba(46, 49, 49, 1)", opacity = 0.2))

produces the following plot, where the arguments color and opacity can be adjusted to your preferred styling. We used the data.frame final_data , since that is where the points are. The variabel year_month was already set by yourself, so there is no additional data wrangling required. To actually generate the dots, be sure to set mode = "markers" in the add_trace() function.