Upload 2 files

.gitattributes

@@ -57,3 +57,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+complete_meta_data.csv filter=lfs diff=lfs merge=lfs -text

complete_meta_data.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04daadece427428d0d659f3896d5a06ba88a8b626edde7c6e54b59251393f33f
+size 213055926

stock_metadata.Rmd

@@ -0,0 +1,566 @@
+---
+title: "Copper"
+output: html_document
+date: "2025-06-23"
+---
+
+```{r setup, include=FALSE}
+knitr::opts_chunk$set(echo = TRUE)
+library(readr)
+library(tidyverse)
+library(readxl)
+library(ggplot2)
+library(dplyr)
+library(httr)
+library(jsonlite)
+```
+
+```{r}
+# 
+# res <- GET("https://eodhd.com/api/screener",
+#            query = list(
+#              industries = "Metals & Mining",
+#              query = "copper",
+#              api_token = " 5f3afd582bd7b4.95720069"
+#            ))
+# 
+# companies <- fromJSON(content(res, "text"))
+# 
+# copper = companies$data
+
+```
+
+```{r}
+
+api_token <- "5f3afd582bd7b4.95720069"
+
+```
+
+```{r}
+
+res <- GET("https://eodhd.com/api/exchanges-list", query = list(api_token = api_token))
+exchange_list <- fromJSON(content(res, "text", encoding = "UTF-8"))
+exchange_df <- as.data.frame(exchange_list)
+
+# Step 2: Initialize storage for results
+all_stocks <- list()
+
+# Step 3: Loop over exchange codes and download stocks
+for (exchange in exchange_df$Code) {
+  message("Fetching data for exchange: ", exchange)
+  
+  res <- GET(
+    paste0("https://eodhd.com/api/exchange-symbol-list/", exchange),
+    query = list(api_token = api_token)
+  )
+  
+  if (status_code(res) == 200) {
+    json_text <- content(res, "text", encoding = "UTF-8")
+    stock_df <- tryCatch(
+      {
+        read.csv(text = json_text)
+      },
+      error = function(e) {
+        message("Skipping ", exchange, " due to error: ", e$message)
+        return(NULL)
+      }
+    )
+    
+    if (!is.null(stock_df) && nrow(stock_df) > 0) {
+      stock_df$ExchangeCode <- exchange
+      all_stocks[[length(all_stocks) + 1]] <- stock_df
+    }
+  } else {
+    message("Failed to fetch data for ", exchange, " (", status_code(res), ")")
+  }
+  
+  Sys.sleep(1)  # avoid rate limit
+}
+
+# Step 3.5: Force all columns to character for binding
+all_stocks_clean <- lapply(all_stocks, function(df) {
+  if (is.data.frame(df)) {
+    df[] <- lapply(df, as.character)  # convert every column to character
+    return(df)
+  } else {
+    return(NULL)
+  }
+})
+all_stocks_clean <- Filter(Negate(is.null), all_stocks_clean)  # remove NULLs
+
+
+# Step 4: Combine cleaned data frames
+stock_metadata_df <- dplyr::bind_rows(all_stocks_clean)
+
+# Optional: Save or view
+View(stock_metadata_df)
+# write.csv(stock_metadata_df, "global_stock_metadata.csv", row.names = FALSE)
+```
+
+# Stock MetaData
+
+```{r}
+merge_fields_deep <- function(base_row, ...) {
+  lists <- list(...)
+  for (list_item in lists) {
+    if (is.null(list_item)) next
+    for (field in names(list_item)) {
+      value <- list_item[[field]]
+      if (is.list(value)) {
+        value <- paste(unlist(value), collapse = ", ")
+      }
+      base_row[[field]] <- value
+    }
+  }
+  return(base_row)
+}
+
+extract_institution_fields <- function(holder_list) {
+  if (is.null(holder_list$Institutions)) return(list())
+  
+  holders <- holder_list$Institutions
+  fields <- list()
+  
+  for (i in seq_along(holders)) {
+    h <- holders[[i]]
+    prefix <- paste0("Holder_", i, "_")
+    fields[[paste0(prefix, "name")]] <- h$name
+    fields[[paste0(prefix, "date")]] <- h$date
+    fields[[paste0(prefix, "totalShares")]] <- h$totalShares
+    fields[[paste0(prefix, "currentShares")]] <- h$currentShares
+    fields[[paste0(prefix, "change_p")]] <- h$change_p
+  }
+  
+  return(fields)
+}
+
+extract_fund_fields <- function(holder_list) {
+  if (is.null(holder_list$Funds)) return(list())
+  
+  funds <- holder_list$Funds
+  fields <- list()
+  
+  for (i in seq_along(funds)) {
+    f <- funds[[i]]
+    prefix <- paste0("Fund_", i, "_")
+    fields[[paste0(prefix, "name")]] <- f$name
+    fields[[paste0(prefix, "date")]] <- f$date
+    fields[[paste0(prefix, "totalShares")]] <- f$totalShares
+    fields[[paste0(prefix, "currentShares")]] <- f$currentShares
+    fields[[paste0(prefix, "change_p")]] <- f$change_p
+  }
+  
+  return(fields)
+}
+
+extract_dividend_fields <- function(div_list) {
+  if (is.null(div_list) || length(div_list) == 0) return(list())
+
+  div_raw <- lapply(div_list, function(x) {
+    if (!is.null(x$Year) && !is.null(x$Count)) {
+      data.frame(Year = as.integer(x$Year), Count = as.integer(x$Count))
+    } else {
+      NULL
+    }
+  })
+
+  div_df <- do.call(rbind, div_raw)
+
+  if (is.null(div_df) || nrow(div_df) == 0) return(list())
+
+  # 保留近5年（最多到 2024）
+  current_year <- as.integer(format(Sys.Date(), "%Y"))
+  target_years <- (current_year - 4):current_year
+
+  div_df <- div_df[div_df$Year %in% target_years, ]
+
+  if (nrow(div_df) == 0) return(list())  # ✅ 加这一句！
+
+  out <- setNames(as.list(div_df$Count), paste0("Dividend_", div_df$Year))
+  return(out)
+
+}
+
+# 主循环开始
+stock_only <- stock_metadata_df %>%
+  filter(Type %in% c("Common Stock", "Preferred Stock", "ETF"))
+
+#stocks_subset <- head(stock_only, 5)
+stocks_subset <- stock_only[94334:96494, ] %>%
+   filter(Type == "Common Stock")
+#stocks_subset$Code <- sprintf("%06d", as.integer(stocks_subset$Code))
+#stocks_subset$Code <- as.character(stocks_subset$Code)
+#stocks_subset <- tail(stocks_subset)
+
+#stocks_subset <- stock_only[50000:nrow(stock_only), ]
+#stocks_subset <- stock_only
+enriched_data <- list()
+
+for (i in 1:nrow(stocks_subset)) {
+  symbol <- as.character(stocks_subset[i, "Code"])
+  exchange <- as.character(stocks_subset[i, "ExchangeCode"])
+  full_symbol <- paste0(symbol, ".", gsub(" ", "", exchange))
+  
+  message(sprintf("🔄 Processing %d / %d: %s", i, nrow(stocks_subset), full_symbol))
+  
+  url <- paste0("https://eodhd.com/api/fundamentals/", full_symbol,
+                "?api_token=", api_token)
+  
+  res <- tryCatch(GET(url), error = function(e) NULL)
+  
+  if (!is.null(res) && status_code(res) == 200) {
+    json_data <- tryCatch(fromJSON(content(res, "text", encoding = "UTF-8")), error = function(e) NULL)
+    
+    if (!is.null(json_data)) {
+      # 提取三部分数据
+      inst_fields <- extract_institution_fields(json_data$Holders)
+      fund_fields <- extract_fund_fields(json_data$Holders)
+      dividend_data <- extract_dividend_fields(json_data$SplitsDividends$NumberDividendsByYear)
+      
+      merged_fields <- c(
+        json_data$General,
+        json_data$Highlights,
+        json_data$Valuation,
+        json_data$SharesStats,
+        json_data$Technicals,
+        json_data$SplitsDividends[names(json_data$SplitsDividends) != "NumberDividendsByYear"],
+        dividend_data,
+        inst_fields,
+        fund_fields
+        
+      )
+      
+      enriched_row <- merge_fields_deep(stocks_subset[i, ], merged_fields)
+      enriched_data[[length(enriched_data) + 1]] <- enriched_row
+    } else {
+      message("⚠️ No JSON data for ", full_symbol)
+    }
+  } else {
+    message("❌ Failed request for ", full_symbol)
+  }
+  
+  Sys.sleep(1)  # 避免过快请求
+}
+
+# 合并输出结果
+if (length(enriched_data) > 0) {
+  result_df <- bind_rows(enriched_data)
+} else {
+  warning("No data enriched.")
+  result_df <- data.frame()
+}
+
+#View(result_df)
+#write.csv(result_df, "metadata(1-1w).csv", row.names = FALSE)
+#write.csv(result_df, "metadata(1w-2w).csv", row.names = FALSE)
+#write.csv(result_df, "metadata_Taiwan(94324:96485).csv", row.names = FALSE)
+```
+
+```{r}
+w1 = read_csv("metadata(1-1w).csv")
+w2 = read_csv("metadata(1w-2w).csv")
+w3 = read_csv("metadata(2w-4w).csv")
+w4 = read_csv("metadata(4w-6w).csv")
+w5 = read_csv("metadata(6w-8w).csv")
+w6 = read_csv("metadata(8w-82080).csv")
+w7 = read_csv("metadata(85250-94323).csv")
+w8 = read_csv("metadata(96486-end).csv")
+w9 = read_csv("metadata_China.csv")
+w10 = read_csv("metadata_Taiwan.csv")
+
+x1 = w1 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x2 = w2 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x3 = w3 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x4 = w4 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x5 = w5 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x6 = w6 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x7 = w7 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x8 = w8 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x9 = w9 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+x10 = w10 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")%>%
+  summarize(total_count = n())
+
+```
+
+```{r}
+yy = rbind(x1,x2,x3,x4,x5,x6,x7,x8,x9,x10)
+yy = yy %>%
+  group_by(Industry) %>%
+  summarize(count = sum(total_count)) %>%
+  arrange(desc(count))
+
+write.csv(yy, "Industry_Count.csv", row.names = FALSE)
+```
+
+```{r}
+x1 = w1 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x2 = w2 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x3 = w3 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x4 = w4 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x5 = w5 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x6 = w6 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x7 = w7 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x8 = w8 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x9 = w9 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+x10 = w10 %>% group_by(Industry) %>%
+  filter(Type == "Common Stock")
+
+
+
+# write.csv(x1, "metadata_1.csv", row.names = FALSE)
+# write.csv(x2, "metadata_2.csv", row.names = FALSE)
+# write.csv(x3, "metadata_3.csv", row.names = FALSE)
+# write.csv(x4, "metadata_4.csv", row.names = FALSE)
+# write.csv(x5, "metadata_5.csv", row.names = FALSE)
+# write.csv(x6, "metadata_6.csv", row.names = FALSE)
+# write.csv(x7, "metadata_7.csv", row.names = FALSE)
+# write.csv(x8, "metadata_8.csv", row.names = FALSE)
+# write.csv(x9, "metadata_9.csv", row.names = FALSE)
+# write.csv(x10, "metadata_10.csv", row.names = FALSE)
+
+```
+
+## Clean Data
+
+```{r}
+extract_top3_holders_clean <- function(df) {
+  # Step 1: 找出 totalShares 列
+  share_cols <- grep("^Holder_\\d+_totalShares$", names(df), value = TRUE)
+  holder_fields <- c("_name", "_date", "_totalShares", "_currentShares")
+  
+  # Step 2: 找出每行中最大的 3 个 holder
+  top3_indices <- apply(df[ , share_cols, drop = FALSE], 1, function(row) {
+    non_na <- which(!is.na(row))
+    if (length(non_na) == 0) return(rep(NA, 3))
+    top <- order(row[non_na], decreasing = TRUE)[1:min(3, length(non_na))]
+    return(non_na[top])
+  })
+  
+  # Step 3: 抽出列名
+  top3_colnames <- lapply(top3_indices, function(idxs) {
+    if (all(is.na(idxs))) return(rep(NA, 3))
+    return(share_cols[idxs])
+  })
+  top3_colnames <- as.data.frame(do.call(rbind, top3_colnames), stringsAsFactors = FALSE)
+  
+  # Step 4: 抽取每行 holder 对应数据并重命名
+  row_extracts <- lapply(1:nrow(top3_colnames), function(i) {
+    bases <- top3_colnames[i, ]
+    if (all(is.na(bases))) {
+      empty_names <- unlist(lapply(1:3, function(j) paste0("Holder_", j, holder_fields)))
+      return(as.data.frame(matrix(NA, nrow = 1, ncol = length(empty_names),
+                                  dimnames = list(NULL, empty_names))))
+    }
+    bases <- bases[!is.na(bases)]
+    cols <- unlist(lapply(bases, function(base) paste0(gsub("_totalShares$", "", base), holder_fields)))
+    out <- df[i, cols, drop = FALSE]
+    new_names <- unlist(lapply(seq_along(bases), function(j) paste0("Holder_", j, holder_fields)))
+    names(out) <- new_names
+    full_names <- unlist(lapply(1:3, function(j) paste0("Holder_", j, holder_fields)))
+    for (nm in setdiff(full_names, names(out))) {
+      out[[nm]] <- NA
+    }
+    out <- out[ , full_names]
+    return(out)
+  })
+  
+  top3_df <- do.call(rbind, row_extracts)
+  
+  # Step 5: 删除原始 holder/fund 列并合并
+  df <- df[ , !grepl("^Holder_|^Fund_", names(df))]
+  df <- cbind(df, top3_df)
+  
+  # Step 6: 删除无关字段
+  filter_out <- c(
+    "CurrencyCode", "CurrencyName", "CurrencySymbol", "CountryISO", "ISIN", "LEI", "Listings", "Officers", "LogoURL",
+    "ShortPercent", "ForwardAnnualDividendRate", "ForwardAnnualDividendYield", "DividendDate", "ExDividendDate",
+    "LastSplitFactor", "LastSplitDate", "Dividend_2021", "Dividend_2022", "Dividend_2023", "Dividend_2024", "Dividend_2025","DelistedDate","Category"
+  )
+  df <- df %>% select(-any_of(filter_out))
+  
+  return(df)
+}
+
+```
+
+```{r}
+
+z1 = extract_top3_holders_clean(x1)
+z2 = extract_top3_holders_clean(x2)
+z3 = extract_top3_holders_clean(x3)
+#z5 = extract_top3_holders_clean(x5)
+z7 = extract_top3_holders_clean(x7)
+
+filter_out <- c(
+    "CurrencyCode", "CurrencyName", "CurrencySymbol", "CountryISO", "ISIN", "LEI", "Listings", "Officers", "LogoURL",
+    "ShortPercent", "ForwardAnnualDividendRate", "ForwardAnnualDividendYield", "DividendDate", "ExDividendDate",
+    "LastSplitFactor", "LastSplitDate", "Dividend_2021", "Dividend_2022", "Dividend_2023", "Dividend_2024", "Dividend_2025", "DelistedDate","Category"
+  )
+rm(list = ls(pattern = "^Fund_"))
+x5 <- x5[ , !grepl("^Fund_", names(x5))]
+rm(list = ls(pattern = "^Holder_"))
+x5 <- x5[ , !grepl("^Holder_", names(x5))]
+z5 = x5%>%
+  select(-any_of(filter_out))
+z4 = x4 %>%
+  select(-any_of(filter_out))
+z6 = x6 %>%
+  select(-any_of(filter_out))
+z8 = x8 %>%
+  select(-any_of(filter_out))
+z9 = x9 %>%
+  select(-any_of(filter_out))
+z10 = x10 %>%
+  select(-any_of(filter_out))
+```
+
+```{r}
+# 需要补齐的 Holder 列名
+holder_fields <- c("_name", "_date", "_totalShares", "_currentShares")
+required_holder_cols <- unlist(lapply(1:3, function(j) paste0("Holder_", j, holder_fields)))
+
+# 获取 x1 到 x10 的数据框
+data_list <- mget(paste0("z", 1:10))
+
+# 用 lapply 批量补全
+data_list_fixed <- lapply(data_list, function(df) {
+  missing_cols <- setdiff(required_holder_cols, names(df))
+  for (col in missing_cols) {
+    df[[col]] <- NA
+  }
+  return(df)
+})
+
+# 还原到环境中（x1 到 x10 被更新）
+list2env(data_list_fixed, .GlobalEnv)
+
+
+date_cols <- grep("_date$", names(z1), value = TRUE)
+
+# 强制转换 z1/z2 的日期列为 Date 类型
+for (col in date_cols) {
+  if (col %in% names(z1)) z1[[col]] <- as.Date(z1[[col]])
+  if (col %in% names(z2)) z2[[col]] <- as.Date(z2[[col]])
+  if (col %in% names(z3)) z3[[col]] <- as.Date(z3[[col]])
+  if (col %in% names(z4)) z4[[col]] <- as.Date(z4[[col]])
+  if (col %in% names(z5)) z5[[col]] <- as.Date(z5[[col]])
+  if (col %in% names(z6)) z6[[col]] <- as.Date(z6[[col]])
+  if (col %in% names(z7)) z7[[col]] <- as.Date(z7[[col]])
+  if (col %in% names(z8)) z8[[col]] <- as.Date(z8[[col]])
+  if (col %in% names(z9)) z9[[col]] <- as.Date(z9[[col]])
+  if (col %in% names(z10)) z10[[col]] <- as.Date(z10[[col]])
+}
+
+
+stock_meta_data <- rbind(z1, z2,z3,z4,z5,z6,z7,z8,z9,z10)
+```
+
+# Filter Industry
+
+```{r}
+ind = c("Other Industrial Metals & Mining", "Gold", "Oil & Gas E&P", "Other Precious Metals & Mining", 
+        "Semiconductors", "Oil & Gas Equipment & Services", "Semiconductor Equipment & Materials", 
+        "Metal Fabrication", "Oil & Gas Refining & Marketing", "Copper", "Oil & Gas Midstream", 
+        "Thermal Coal", "Oil & Gas Integrated", "Uranium", "Silver", "Oil & Gas Drilling", "Metals & Mining",
+        "Oil & Gas", "Coal", "Oil & Gas Related Equipment and Services", "Oil, Gas & Consumable Fuels"
+        )
+
+cleaned_data = stock_meta_data %>%
+  filter(Industry %in% ind)
+write.csv(cleaned_data, "cleaned_stock_data.csv", row.names = FALSE)
+```
+
+# Merge Financial
+
+```{r}
+fin = read.csv("financials_filtered.csv")
+stock_meta_data$PrimaryTicker <- paste0(stock_meta_data$Code, ".", stock_meta_data$ExchangeCode)
+
+merged = stock_meta_data %>%
+  left_join(fin, by = c("PrimaryTicker" = "Ticker")) 
+
+merged_industry = merged %>%
+  filter(Industry %in% ind)
+```
+
+# EDA
+
+```{r}
+# 选择相关列
+equity_cols <- paste0("totalStockholderEquity_", 2020:2024)
+df_equity <- merged_industry[, c("PrimaryTicker", equity_cols)]
+
+# 定义 CAGR 函数
+calc_cagr <- function(start, end, years = 4) {
+  if (is.na(start) || is.na(end) || start <= 0) return(NA)
+  return((end / start)^(1 / years) - 1)
+}
+
+# 逐行计算 CAGR（2020~2024）
+df_equity$Equity_CAGR_2020_2024 <- mapply(
+  calc_cagr,
+  df_equity$totalStockholderEquity_2020,
+  df_equity$totalStockholderEquity_2024
+)
+
+```
+
+```{r}
+library(tidyverse)
+
+# 假设 df_equity 已包含 Ticker, totalStockholderEquity_2020~2024 和 CAGR 列
+# 选出 CAGR 最大的前 5 家公司
+top5 <- df_equity %>%
+  arrange(desc(Equity_CAGR_2020_2024)) %>%
+  slice(1:5)
+
+# 把数据转成长格式，适合 ggplot2 绘图
+df_long <- top5 %>%
+  select(PrimaryTicker, starts_with("totalStockholderEquity_")) %>%
+  pivot_longer(
+    cols = -PrimaryTicker,
+    names_to = "Year",
+    values_to = "Equity"
+  ) %>%
+  mutate(Year = as.numeric(gsub("totalStockholderEquity_", "", Year)))
+
+# 画折线图
+ggplot(df_long, aes(x = Year, y = Equity, color = PrimaryTicker)) +
+  geom_line(size = 1) +
+  geom_point(size = 2) +
+  labs(
+    title = "Top 5 Companies by Equity CAGR (2020–2024)",
+    x = "Year", y = "Total Stockholder Equity"
+  ) +
+  theme_minimal() +
+  scale_y_continuous(labels = scales::comma)
+
+```