stock_visualization.Rmd

```{r setup, include=FALSE}
library(knitr)
knitr::opts_chunk$set(echo = TRUE)
```

```{r data input}
setwd("/Users/Alesa/Documents/Kaisa/Data_aquisition/stock_data_processed3");
my_files <- list.files("/Users/Alesa/Documents/Kaisa/Data_aquisition/stock_data_processed3"); # Please download the stock data and load it locally.
stock_prc <- lapply(my_files, read.csv) # Note: stock_prc[[i]] is a data frame type
logrets <- lapply(stock_prc, function(x) diff(data.matrix(log(x[6:nrow(x), 2:ncol(x)])), lag=1)) # Note: logrets[[i]] is a matrix type.
my_files<-gsub(".csv", "", list.files("/Users/Alesa/Documents/Kaisa/Data_aquisition/stock_data_processed3", full.names = FALSE), fixed = TRUE);
corr_mat <- lapply(logrets, cor, method = c("pearson")) # Note: corr_mat[[i]] is a matrix type.
dist_mat <- lapply(corr_mat, function(x) sqrt(2 * (1 - x)))
```

```{r generate clusters and create adjacency matrix}
library("VisualStock")
trial_dist <- dist_mat[[66]] # Choose stocks within a period!
cluster_output<-stock_cluster(trial_dist)
```

```{r generate parameters for plotting}
gic_sec <- seq(10, 60, by = 5)# 11 sectors
gic_name <- c("Energy", "Materials", "Industrials", "Consumer Discretionary", "Consumer Staples", "Health Care", "Financials", "Information Technology", "Communication Service", "Utilities", "Real Estate")
l <- cluster_output$num_vertices # find total number of vertices of the adjacency matrix

Mode_Max <- function(x) {
  ux <- unique(x)
  sec_no <- ux[which.max(tabulate(match(x, ux)))]
  output <- as.integer(sec_no/5 -1) # This line is model-related, must convert the group info to a group of number within 1,2,3,...N (to match the color number provided by d3.js).
} # find the first most frequently appearing term in (vector) x
Mode_All <- function(x){
  counts <- as.data.frame(table(x))
  colnames(counts) <- c("sec", "freq")
  counts <- merge(data.frame(sec = gic_sec), counts, all.x = TRUE) # note: be careful, takes global value
  counts[is.na(counts)] <- 0
  output <- as.integer(counts[, 2])  
} # count the frequency of sectors in each vertex

vertex.group <- as.integer(rep(0, l)) # most frequently appearing sector number in each vertex
vertex.size <- as.integer(rep(0, l)) # number of points in each vertex
vertex.comp <- list() # count number of different sectors in each vertex

for (i in 1:l){
   points.in.vertex <- cluster_output$points_in_vertex[[i]]
   sectors.in.vertex <- as.integer(stock_prc[[66]][2, points.in.vertex + 1])
   vertex.group[i] <- Mode_Max(sectors.in.vertex)
   vertex.size[i] <- length(points.in.vertex)
   vertex.comp[[i]] <- Mode_All(sectors.in.vertex)
}
```

```{r igraph 2D pie chart plot, echo = FALSE}
library(igraph)
stock.graph <- graph_from_adjacency_matrix(cluster_output$adjacency, mode="undirected", weighted=TRUE, diag=FALSE)
plot(stock.graph, layout = layout_in_circle, edge.width = E(stock.graph)$weight, main = my_files[66], margin = 0)

attach(mtcars)
layout(matrix(c(1, 1, 2, 2), 1, 4, byrow = TRUE))
plot(stock.graph, layout = layout_in_circle, vertex.shape = "pie", vertex.pie = vertex.comp, vertex.pie.color = list(rainbow(11)), vertex.size = 10*vertex.size^0.2, main = my_files[66], margin = 0)
pie(rep(1,11), labels = paste0(gic_sec, "  ", gic_name), clockwise = TRUE, col = rainbow(11),  radius = 0.5, cex = 0.9, main = 'Graph Legend')

```

```{r generate json file for 3D interactive plot}
library("VisualStock")
net <- igraph_to_net(stock.graph, group = vertex.group, size = vertex.size^0.2, comp = vertex.comp) # normalize the radius of the node so that the last one is not too big.
net_to_json(net, "/Users/Alesa/Downloads/pied3/trial.json")
```