first commit

.Rhistory

@@ -0,0 +1,121 @@
+install.packages("rvest")
+url <- "https://en.wikipedia.org/wiki/Roy_Williams_(coach)#Head_coaching_record"
+library(tidyverse)
+library(rvest)
+rw <- url %>% html() %>% html_nodes(xpath = '//*[@id="mw-content-text"]/table[3]') %>% html_table()
+rw <- url %>% read_html() %>%
+html_nodes(xpath = '//*[@id="mw-content-text"]/table[3]') %>%
+html_table()
+rw <- url %>% read_html() %>%
+html_nodes(xpath = '//*[@id="mw-content-text"]/table[3]', fill = TRUE) %>%
+html_table()
+rw <- url %>% read_html() %>%
+html_nodes(xpath = '//*[@id="mw-content-text"]/table[3]') %>%
+html_table(fill = TRUE)
+rw
+str(rw)
+rw$Overall
+rw$" Overall"
+rw[[1]]$Overall
+rw_records <- rw[[1]]$Overall
+head(rw_records)
+rw_year <- rw[[1]]$Season
+rw_year
+df_rw <- data.frame(rw_year, rw_records)
+head(df_rw)
+nchar(df_rw)
+nchar(df_rw[1,1])
+nchar(df_rw$rw_year)
+nchar(df_rw$rw_year[[1]])
+df_year[[1]]
+rw_year[[1]]
+nchar(rw_year[[1]])
+str(df_rw)
+df_rw <- tbl(rw_year, rw_records)
+df_rw <- tibble(rw_year, rw_records)
+str(df_rw)
+df_rw <- df_rw %>% filter(nchar(rw_year) < 10)
+head(df_rw)
+library(attutilr)
+headtail(dr_rw)
+headtail(df_rw)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+dim(df_rw)
+headtail(df_rw)
+grepl(df_rw$rw_records[1],"-")
+df_rw$rw_records[1]
+grepl(df_rw$rw_records[1],"–")
+?grepl
+grepl("–", df_rw$rw_records[1])
+grep("–", df_rw$rw_records[1])
+?strsplit
+df_rw <- df_rw %>% mutate(win = strsplit(rw_records, "–")[1], loss = strsplit(rw_records, "–")[2])
+strsplit(df_rw$rw_records[1])[1]
+strsplit(df_rw$rw_records[1], "–")[1]
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% mutate(win = strsplit(rw_records, "–")[1], loss = strsplit(rw_records, "–")[2])
+library(stringr)
+?str_split
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% mutate(win = str_split("–", rw_records)[1], loss = str_split("–", rw_records)[2])
+str_split("–", df_rw$rw_records[1])
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+str_split("–", df_rw$rw_records[1])
+str_split("–", "10-20")
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% mutate(win = str_split(rw_records, "–", )[1], loss = str_split( rw_records, "–")[2])
+str_split(10-20", "–")
+str_split("10–20", "–")
+df_rw <- df_rw %>% mutate(win = str_split(rw_records, "–")[1], loss = str_split( rw_records, "–")[2])
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% mutate(win = str_split(rw_records, "–")[1], loss = str_split( rw_records, "–")[2])
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+headtail(df_rw)
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% mutate(win = str_split(rw_records, "–")[[1]], loss = str_split( rw_records, "–")[[2]])
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% mutate(win = str_split(rw_records, "–")[[1]][[1]], loss = str_split( rw_records, "–")[[1]][[2]])
+headtail(df_rw)
+str_split(df_rw$rw_records, "–")
+str(df_rw)
+?split
+?unite
+?separate
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+#df_rw <- df_rw %>% mutate(win = str_split(rw_records, "–")[[1]][[1]], loss = str_split( rw_records, "–")[[1]][[2]])
+df_rw <- df_rw %>% separate(rw_records, c("win", "loss"))
+headtail(df_Rw)
+headtail(df_rw)
+sum(df_rw$win)
+df_rw <- df_rw %>% separate(rw_records, c("win", "loss"), convert = TRUE)
+rw_year <- rw[[1]]$Season
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% separate(rw_records, c("win", "loss"), convert = TRUE)
+sum(df_rw$win)
+str(df_rw$win)
+str(df_rw)
+df_rw <- df_rw %>% mutate(career_yr = row_index())
+?row_index
+?index_row
+df_rw <- df_rw %>% mutate(career_yr = row_number())
+headtail(df_rw)
+str(df_rw)
+df_rw %>% ggplot(aes(x = carrer_yr, y = win)) + geom_point()
+df_rw %>% ggplot(aes(x = career_yr, y = win)) + geom_point()
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% separate(rw_records, c("win", "loss"), convert = TRUE)
+df_rw <- df_rw %>% mutate(win_cum = cumsum(win), career_yr = row_number())
+df_rw %>% ggplot(aes(x = career_yr, y = win)) + geom_point()
+df_rw %>% ggplot(aes(x = career_yr, y = win_cum)) + geom_point()

.Rproj.user/BFAA1C1E/pcs/files-pane.pper

@@ -0,0 +1,9 @@
+{
+    "path" : "~/projects/coaches",
+    "sortOrder" : [
+        {
+            "ascending" : true,
+            "columnIndex" : 2
+        }
+    ]
+}

.Rproj.user/BFAA1C1E/pcs/source-pane.pper

@@ -0,0 +1,3 @@
+{
+    "activeTab" : 2
+}

.Rproj.user/BFAA1C1E/pcs/windowlayoutstate.pper

@@ -0,0 +1,14 @@
+{
+    "left" : {
+        "panelheight" : 665,
+        "splitterpos" : 283,
+        "topwindowstate" : "NORMAL",
+        "windowheight" : 704
+    },
+    "right" : {
+        "panelheight" : 665,
+        "splitterpos" : 424,
+        "topwindowstate" : "NORMAL",
+        "windowheight" : 704
+    }
+}

.Rproj.user/BFAA1C1E/pcs/workbench-pane.pper

@@ -0,0 +1,6 @@
+{
+    "TabSet1" : 0,
+    "TabSet2" : 1,
+    "TabZoom" : {
+    }
+}

.Rproj.user/BFAA1C1E/rmd-outputs

@@ -0,0 +1,5 @@
+
+
+
+
+

.Rproj.user/BFAA1C1E/saved_source_markers

@@ -0,0 +1 @@
+{"active_set":"","sets":[]}

.Rproj.user/BFAA1C1E/sdb/prop/A5126009

@@ -0,0 +1,3 @@
+{
+    "tempName" : "Untitled1"
+}

.Rproj.user/BFAA1C1E/sdb/prop/BD41F207

@@ -0,0 +1,3 @@
+{
+    "tempName" : "Untitled1"
+}

.Rproj.user/BFAA1C1E/sdb/prop/F2CB869

@@ -0,0 +1,3 @@
+{
+    "tempName" : "Untitled1"
+}

.Rproj.user/BFAA1C1E/sdb/prop/INDEX

@@ -0,0 +1,3 @@
+~%2Fprojects%2Fcoaches%2Fmain.R="BD41F207"
+~%2Fprojects%2Fcoaches%2Fscratch.R="F2CB869"
+~%2Fprojects%2Fcoaches%2Futils.R="A5126009"

.Rproj.user/BFAA1C1E/sdb/s-5FD54A97/4F786E09

@@ -0,0 +1,21 @@
+{
+    "collab_server" : "",
+    "contents" : "\ngrab_records <- function(src){\n  \n  coach <- src[[\"URL\"]] %>% read_html() %>% html_nodes(xpath = src[[\"Table\"]]) %>% html_table(fill = TRUE)\n  records <- coach[[1]]$Overall\n  year <- coach[[1]]$Season\n  df_coach <- tibble(year, records)\n  df_coach <- df_coach %>% filter(nchar(records) < 6)\n  df_coach <- df_coach %>% separate(records, c(\"win\", \"loss\"), convert = TRUE)\n  df_coach <- df_coach %>% mutate(win_cum = cumsum(win), career_yr = row_number())\n  \n  df_coach\n}",
+    "created" : 1486136448518.000,
+    "dirty" : false,
+    "encoding" : "UTF-8",
+    "folds" : "",
+    "hash" : "2801794642",
+    "id" : "4F786E09",
+    "lastKnownWriteTime" : 1486146376,
+    "last_content_update" : 1486146376610,
+    "path" : "~/projects/coaches/utils.R",
+    "project_path" : "utils.R",
+    "properties" : {
+        "tempName" : "Untitled1"
+    },
+    "relative_order" : 2,
+    "source_on_save" : false,
+    "source_window" : "",
+    "type" : "r_source"
+}

.Rproj.user/BFAA1C1E/sdb/s-5FD54A97/B4C7229

@@ -0,0 +1,21 @@
+{
+    "collab_server" : "",
+    "contents" : "library(tidyverse)\nlibrary(rvest)\n\nsource(\"utils.R\")\n\n#---------------\n# read data\n#---------------\ncoaches <- read.csv(\"input_data_table.csv\", header = TRUE, as.is = TRUE, quote = \"\")\ncoaches$career_records <- apply(coaches, 1, grab_records)\n# why not working? \n#coaches <- src %>% mutate(career_records = apply(coaches, 1, grab_records))\nsetNames(coaches$career_records, coaches$Coach)\n# latest wins\ncoaches$latest_wins <- coaches$career_records %>% map_int(function(x) x$win_cum[nrow(x)])\n\n#---------------\n# total wins\n#---------------\ncoaches %>% ggplot(aes(x = reorder(Coach, latest_wins), y = latest_wins)) + geom_bar(stat = \"identity\") + coord_flip() + theme_bw()\n\n#---------------\n# win by career year\n#---------------\n#df_career_records <- bind_rows(coaches$career_records, .id = \"Coach\")\n#df_career_records <- rbindlist(coaches$career_records, idcol = \"Coach\")\n\ndf_career_records <- coaches %>% unnest() %>% select(Coach, career_yr, win_cum)\ndf_career_records$type <- ifelse(df_career_records$Coach %in% c(\"Dean Smith\", \"Roy Williams\", \"Coach K\"), TRUE, FALSE)\ndf_career_records %>% ggplot(aes(x = career_yr, y = win_cum)) + geom_point(aes(group = Coach, colour = type)) + geom_line(aes(group = Coach, colour = type)) + theme_bw() + stat_smooth()\n\n",
+    "created" : 1486146195326.000,
+    "dirty" : false,
+    "encoding" : "UTF-8",
+    "folds" : "",
+    "hash" : "291307219",
+    "id" : "B4C7229",
+    "lastKnownWriteTime" : 1486264726,
+    "last_content_update" : 1486264726884,
+    "path" : "~/projects/coaches/main.R",
+    "project_path" : "main.R",
+    "properties" : {
+        "tempName" : "Untitled1"
+    },
+    "relative_order" : 3,
+    "source_on_save" : false,
+    "source_window" : "",
+    "type" : "r_source"
+}

.Rproj.user/BFAA1C1E/sdb/s-5FD54A97/BC60C9A7

@@ -0,0 +1,21 @@
+{
+    "collab_server" : "",
+    "contents" : "install.packages(\"rvest\")\nurl <- \"https://en.wikipedia.org/wiki/Roy_Williams_(coach)#Head_coaching_record\"\n\nlibrary(tidyverse)\nlibrary(rvest)\n\nrw <- url %>% read_html() %>% \n  html_nodes(xpath = \"//*[@id=\\\"mw-content-text\\\"]/table[3]\") %>% html_table(fill = TRUE)\n\nrw_records <- rw[[1]]$Overall\nrw_year <- rw[[1]]$Season\ndf_rw <- tibble(rw_year, rw_records)\ndf_rw <- df_rw %>% filter(nchar(rw_records) < 6)\ndf_rw <- df_rw %>% separate(rw_records, c(\"win\", \"loss\"), convert = TRUE)\ndf_rw <- df_rw %>% mutate(win_cum = cumsum(win), career_yr = row_number())\n\ndf_rw %>% ggplot(aes(x = career_yr, y = win_cum)) + geom_point()\n\nsrc <- read.csv(\"input_data_table.csv\", header = TRUE, as.is = TRUE, quote = \"\")\ncoach <- src[2, \"URL\"] %>% read_html() %>% html_nodes(xpath = src[2, \"Table\"]) %>% html_table(fill = TRUE)\nrecords <- coach[[1]]$Overall\nyear <- coach[[1]]$Season\ndf_coach <- tibble(year, records)\ndf_coach <- df_coach %>% filter(nchar(records) < 6)\ndf_coach <- df_coach %>% separate(records, c(\"win\", \"loss\"), convert = TRUE)\ndf_coach <- df_coach %>% mutate(win_cum = cumsum(win), career_yr = row_number())\n\nsrc$career_records <- apply(src, 1, grab_records)\n\ncoaches <- read.csv(\"input_data_table.csv\", header = TRUE, as.is = TRUE, quote = \"\")\ncoaches$career_records <- apply(coaches, 1, grab_records)\ncoaches$latest_win_total <- \n  \nmap_int(coaches$carrer_records, win_cum)\n",
+    "created" : 1486134762674.000,
+    "dirty" : false,
+    "encoding" : "UTF-8",
+    "folds" : "",
+    "hash" : "2877864819",
+    "id" : "BC60C9A7",
+    "lastKnownWriteTime" : 1486155240,
+    "last_content_update" : 1486155240371,
+    "path" : "~/projects/coaches/scratch.R",
+    "project_path" : "scratch.R",
+    "properties" : {
+        "tempName" : "Untitled1"
+    },
+    "relative_order" : 1,
+    "source_on_save" : false,
+    "source_window" : "",
+    "type" : "r_source"
+}

.Rproj.user/BFAA1C1E/session-persistent-state

@@ -0,0 +1 @@
+virtual-session-id="8B7A30FB"

.Rproj.user/shared/notebooks/paths

@@ -0,0 +1,3 @@
+/Users/jl939a/projects/coaches/main.R="20F3A554"
+/Users/jl939a/projects/coaches/scratch.R="3BE5A5D5"
+/Users/jl939a/projects/coaches/utils.R="9C8794B9"

coaches.Rproj

@@ -0,0 +1,13 @@
+Version: 1.0
+
+RestoreWorkspace: Default
+SaveWorkspace: Default
+AlwaysSaveHistory: Default
+
+EnableCodeIndexing: Yes
+UseSpacesForTab: Yes
+NumSpacesForTab: 2
+Encoding: UTF-8
+
+RnwWeave: Sweave
+LaTeX: pdfLaTeX

input_data_table.csv

@@ -0,0 +1,13 @@
+Number,Coach,URL,Table
+1,Roy Williams,https://en.wikipedia.org/wiki/Roy_Williams_(coach)#Head_coaching_record,//*[@id="mw-content-text"]/table[3]
+2,Coach K,https://en.wikipedia.org/wiki/Mike_Krzyzewski,//*[@id="mw-content-text"]/table[2]
+3,Bob Knight,https://en.wikipedia.org/wiki/Bob_Knight,//*[@id="mw-content-text"]/table[2]
+4,John Calipari,https://en.wikipedia.org/wiki/John_Calipari,//*[@id="mw-content-text"]/table[3]
+5,Dean Smith,https://en.wikipedia.org/wiki/Dean_Smith,//*[@id="mw-content-text"]/table[2]
+6,Jim Hoeheim,https://en.wikipedia.org/wiki/Jim_Boeheim,//*[@id="mw-content-text"]/table[2]
+7,Adolph Rupp,https://en.wikipedia.org/wiki/Adolph_Rupp,//*[@id="mw-content-text"]/table[3]
+8,Jim Calhoun,https://en.wikipedia.org/wiki/Jim_Calhoun,//*[@id="mw-content-text"]/table[2]
+9,Lute Olson,https://en.wikipedia.org/wiki/Lute_Olson,//*[@id="mw-content-text"]/table[3]
+10,Pat Summitt,https://en.wikipedia.org/wiki/Pat_Summitt,//*[@id="mw-content-text"]/table[5]
+
+

main.R

@@ -0,0 +1,31 @@
+library(tidyverse)
+library(rvest)
+
+source("utils.R")
+
+#---------------
+# read data
+#---------------
+coaches <- read.csv("input_data_table.csv", header = TRUE, as.is = TRUE, quote = "")
+coaches$career_records <- apply(coaches, 1, grab_records)
+# why not working? 
+#coaches <- src %>% mutate(career_records = apply(coaches, 1, grab_records))
+setNames(coaches$career_records, coaches$Coach)
+# latest wins
+coaches$latest_wins <- coaches$career_records %>% map_int(function(x) x$win_cum[nrow(x)])
+
+#---------------
+# total wins
+#---------------
+coaches %>% ggplot(aes(x = reorder(Coach, latest_wins), y = latest_wins)) + geom_bar(stat = "identity") + coord_flip() + theme_bw()
+
+#---------------
+# win by career year
+#---------------
+#df_career_records <- bind_rows(coaches$career_records, .id = "Coach")
+#df_career_records <- rbindlist(coaches$career_records, idcol = "Coach")
+
+df_career_records <- coaches %>% unnest() %>% select(Coach, career_yr, win_cum)
+df_career_records$type <- ifelse(df_career_records$Coach %in% c("Dean Smith", "Roy Williams", "Coach K"), TRUE, FALSE)
+df_career_records %>% ggplot(aes(x = career_yr, y = win_cum)) + geom_point(aes(group = Coach, colour = type)) + geom_line(aes(group = Coach, colour = type)) + theme_bw() + stat_smooth()
+

scratch.R

@@ -0,0 +1,34 @@
+install.packages("rvest")
+url <- "https://en.wikipedia.org/wiki/Roy_Williams_(coach)#Head_coaching_record"
+
+library(tidyverse)
+library(rvest)
+
+rw <- url %>% read_html() %>% 
+  html_nodes(xpath = "//*[@id=\"mw-content-text\"]/table[3]") %>% html_table(fill = TRUE)
+
+rw_records <- rw[[1]]$Overall
+rw_year <- rw[[1]]$Season
+df_rw <- tibble(rw_year, rw_records)
+df_rw <- df_rw %>% filter(nchar(rw_records) < 6)
+df_rw <- df_rw %>% separate(rw_records, c("win", "loss"), convert = TRUE)
+df_rw <- df_rw %>% mutate(win_cum = cumsum(win), career_yr = row_number())
+
+df_rw %>% ggplot(aes(x = career_yr, y = win_cum)) + geom_point()
+
+src <- read.csv("input_data_table.csv", header = TRUE, as.is = TRUE, quote = "")
+coach <- src[2, "URL"] %>% read_html() %>% html_nodes(xpath = src[2, "Table"]) %>% html_table(fill = TRUE)
+records <- coach[[1]]$Overall
+year <- coach[[1]]$Season
+df_coach <- tibble(year, records)
+df_coach <- df_coach %>% filter(nchar(records) < 6)
+df_coach <- df_coach %>% separate(records, c("win", "loss"), convert = TRUE)
+df_coach <- df_coach %>% mutate(win_cum = cumsum(win), career_yr = row_number())
+
+src$career_records <- apply(src, 1, grab_records)
+
+coaches <- read.csv("input_data_table.csv", header = TRUE, as.is = TRUE, quote = "")
+coaches$career_records <- apply(coaches, 1, grab_records)
+coaches$latest_win_total <- 
+
+map_int(coaches$carrer_records, win_cum)

utils.R

@@ -0,0 +1,13 @@
+
+grab_records <- function(src){
+
+  coach <- src[["URL"]] %>% read_html() %>% html_nodes(xpath = src[["Table"]]) %>% html_table(fill = TRUE)
+  records <- coach[[1]]$Overall
+  year <- coach[[1]]$Season
+  df_coach <- tibble(year, records)
+  df_coach <- df_coach %>% filter(nchar(records) < 6)
+  df_coach <- df_coach %>% separate(records, c("win", "loss"), convert = TRUE)
+  df_coach <- df_coach %>% mutate(win_cum = cumsum(win), career_yr = row_number())
+
+  df_coach
+}