shokai / Collective Intelligence study / changeset / 3fc1fa966059

Up to file-list 03/bicluster.rb:

@@ -14,7 +14,7 @@ class Bicluster
   end
   def vec
-    return vec
+    return @vec
   end
   def left

Up to file-list 03/clusters.rb:

@@ -22,9 +22,7 @@ class Clusters
     while clust.length > 1
       lowestpair = [0,1]
       closest = self.method(distance).call(clust[0].vec, clust[1].vec)
       puts 'asdf'
+      p closest
       # すべての組をループし、もっとも距離の近い組を探す
       for i in 0...clust.length
         for j in i+1...clust.length
@@ -33,13 +31,33 @@ class Clusters
             distances[clust[i].id, clust[j].id] = self.method(distance).call(clust[i].vec, clust[j].vec)
           end
-          d = self.method(distance).call(clust[i].id, clust[j].id)
+          puts 'hoaaa'
           d = [distances[clust[i].id], distances[clust[j].id]]
           if d[0] < closest && d[1] < closest
             closest = d
             lowestpair = [1,j]
           end
         end
       end
       # 2つのクラスタの平均を計算する
       mergevec = Array.new
       for i in 0...clust[0].vec.length
         m = (clust[lowestpair[0]].vec[i] + clust[lowestpair[1]].vec[i])/2.0
         mergevec.push(m)
       end
       # 新たなクラスタを作る
       newcluster = Bicluster.new(mergevec, clust[lowestpair[0]], clust[lowestpair[1]], closest, currentclustid)
       # 元のセットではないクラスタのIDは負にする
       currentclustid -= 1
       clust.delete_at(lowestpair[1])
       clust.delete_at(lowestpair[0])
       clust.push(newcluster)
     end
+    return clust[0]
   end
   # ピアソン相関距離を計算
@@ -98,7 +116,11 @@ class Clusters
       # それぞれの行の最初の列は行の名前（blog名）
       rownames.push(tmp.shift)
       # 行の残りの部分がその行のデータ
-      data.push(tmp)
+      wordcount = Array.new
       tmp.each{ |c|
         wordcount.push(c.to_i)
+      }
       data.push(wordcount)
+    }
     return rownames,colnames,data

Up to file-list 03/log:

 ["designed",
  "",
  "pretty",
  "approach",
  "folks",
  "perfect",
  "below",
  "appear",
  "plan",
  "across",
  "jpg",
  "play",
  "creative",
  "developers",
  "brings",
  "starts",
  "brought",
  "absolutely",
  "tried",
  "book",
  "success",
  "second",
  "similar",
  "costs",
  "street",
  "together",
  "rules",
  "anyone",
  "quite",
  "publishing",
  "working",
  "b",
  "f",
  "g",
  "h",
  "j",
  "k",
  "l",
  "direct",
  "price",
  "n",
  "o",
  "almost",
  "q",
  "r",
  "u",
  "v",
  "w",
  "x",
  "y",
  "z",
  "feedburner",
  "companies",
  "actual",
  "starting",
  "field",
  "next",
  "icon",
  "traditional",
  "behind",
  "tips",
  "maps",
  "isn",
  "lines",
  "middle",
  "wall",
  "solid",
  "presidential",
  "check",
  "mark",
  "wants",
  "gif",
  "blogspot",
  "goes",
  "writing",
  "green",
  "google",
  "needs",
  "scale",
  "biden",
  "bloggers",
  "however",
  "economy",
  "means",
  "links",
  "htm",
  "election",
  "super",
  "god",
  "expected",
  "bunch",
  "credit",
  "gov",
  "including",
  "small",
  "said",
  "course",
  "decided",
  "response",
  "plus",
  "hours",
  "dark",
  "runs",
  "place",
  "nbsp",
  "without",
  "code",
  "data",
  "date",
  "history",
  "reasons",
  "file",
  "police",
  "class",
  "speaking",
  "friday",
  "apparently",
  "audience",
  "visit",
  "guy",
  "false",
  "feeds",
  "fine",
  "smart",
  "living",
  "float",
  "games",
  "little",
  "standard",
  "october",
  "looks",
  "star",
  "weeks",
  "car",
  "stay",
  "london",
  "despite",
  "fire",
  "present",
  "problems",
  "lots",
  "friend",
  "tell",
  "allowscriptaccess",
  "tells",
  "details",
  "love",
  "attention",
  "culture",
  "crisis",
  "brand",
  "party",
  "each",
  "thinking",
  "step",
  "ceo",
  "pheedo",
  "five",
  "block",
  "home",
  "vote",
  "reading",
  "turns",
  "running",
  "created",
  "copy",
  "children",
  "term",
  "process",
  "technology",
  "amazon",
  "says",
  "seems",
  "finally",
  "possible",
  "control",
  "align",
  "blogs",
  "computer",
  "among",
  "sure",
  "ever",
  "especially",
  "eye",
  "series",
  "category",
  "host",
  "polls",
  "color",
  "text",
  "particular",
  "extra",
  "stop",
  "hour",
  "act",
  "office",
  "question",
  "customer",
  "keep",
  "speech",
  "calls",
  "move",
  "quality",
  "clip",
  "age",
  "object",
  "answer",
  "ago",
  "create",
  "dollars",
  "press",
  "education",
  "specific",
  "inside",
  "felt",
  "audio",
  "energy",
  "cut",
  "air",
  "easy",
  "moment",
  "track",
  "nytimes",
  "articles",
  "season",
  "outside",
  "life",
  "platform",
  "helps",
  "knowledge",
  "via",
  "public",
  "wireless",
  "minute",
  "church",
  "candidates",
  "taken",
  "several",
  "picture",
  "places",
  "span",
  "amp",
  "takes",
  "provides",
  "messages",
  "research",
  "changes",
  "larger",
  "talk",
  "points",
  "role",
  "does",
  "sunday",
  "hands",
  "wordpress",
  "light",
  "happy",
  "rather",
  "tag",
  "whole",
  "within",
  "build",
  "ask",
  "built",
  "started",
  "making",
  "line",
  "link",
  "offer",
  "questions",
  "explain",
  "jump",
  "legal",
  "ten",
  "happened",
  "future",
  "believe",
  "able",
  "readers",
  "wrong",
  "majority",
  "single",
  "high",
  "entertainment",
  "release",
  "gadgets",
  "percent",
  "president",
  "live",
  "best",
  "url",
  "social",
  "everyone",
  "sold",
  "washington",
  "human",
  "style",
  "she",
  "short",
  "client",
  "wrote",
  "chris",
  "blogger",
  "music",
  "tom",
  "top",
  "fall",
  "red",
  "around",
  "rel",
  "spend",
  "expect",
  "reviews",
  "mind",
  "wikipedia",
  "spent",
  "three",
  "adding",
  "having",
  "international",
  "fans",
  "obviously",
  "guide",
  "city",
  "global",
  "sort",
  "left",
  "anyway",
  "friends",
  "imagine",
  "early",
  "country",
  "board",
  "month",
  "reader",
  "software",
  "result",
  "easier",
  "record",
  "googleadservices",
  "pay",
  "under",
  "head",
  "pages",
  "strong",
  "hear",
  "community",
  "part",
  "permalink",
  "open",
  "per",
  "product",
  "advertising",
  "north",
  "study",
  "common",
  "rss",
  "stuff",
  "previous",
  "building",
  "auto",
  "feedproxy",
  "run",
  "watch",
  "system",
  "option",
  "lets",
  "local",
  "review",
  "area",
  "related",
  "features",
  "aren",
  "multiple",
  "different",
  "baby",
  "water",
  "profile",
  "devices",
  "wiki",
  "server",
  "border",
  "nice",
  "favorite",
  "doing",
  "matter",
  "once",
  "action",
  "looking",
  "myspace",
  "products",
  "writer",
  "words",
  "aspx",
  "thanks",
  "launch",
  "ones",
  "side",
  "old",
  "net",
  "leading",
  "months",
  "index",
  "wish",
  "consumer",
  "white",
  "makes",
  "during",
  "put",
  "limited",
  "embed",
  "applications",
  "bank",
  "ba",
  "bb",
  "bc",
  "camera",
  "bd",
  "bf",
  "updates",
  "took",
  "tool",
  "org",
  "network",
  "game",
  "br",
  "mentioned",
  "follow",
  "campaign",
  "pictures",
  "da",
  "db",
  "dc",
  "dd",
  "de",
  "either",
  "sarah",
  "df",
  "level",
  "interest",
  "added",
  "sometimes",
  "based",
  "law",
  "states",
  "marketing",
  "editor",
  "fa",
  "wonder",
  "fb",
  "total",
  "fc",
  "fd",
  "fe",
  "phone",
  "ff",
  "hspace",
  "non",
  "fm",
  "sources",
  "fs",
  "against",
  "playing",
  "position",
  "height",
  "static",
  "york",
  "hl",
  "hm",
  "spending",
  "let",
  "basic",
  "ht",
  "dead",
  "write",
  "family",
  "deal",
  "worth",
  "something",
  "voice",
  "else",
  "la",
  "message",
  "li",
  "table",
  "site",
  "lm",
  "late",
  "photo",
  "script",
  "himself",
  "involved",
  "economic",
  "content",
  "call",
  "least",
  "shot",
  "members",
  "deep",
  "show",
  "services",
  "reported",
  "pc",
  "near",
  "experience",
  "javascript",
  "mobile",
  "pm",
  "blockquote",
  "pr",
  "completely",
  "px",
  "whether",
  "black",
  "catch",
  "value",
  "sense",
  "books",
  "saying",
  "notice",
  "size",
  "lot",
  "low",
  "need",
  "quick",
  "sharing",
  "card",
  "td",
  "care",
  "th",
  "food",
  "knew",
  "tv",
  "looked",
  "provide",
  "idea",
  "woman",
  "vm",
  "views",
  "vs",
  "reason",
  "xb",
  "private",
  "hasn",
  "west",
  "instead",
  "difference",
  "appears",
  "break",
  "terms",
  "options",
  "simple",
  "minutes",
  "child",
  "women",
  "zm",
  "students",
  "simply",
  "flash",
  "four",
  "upon",
  "financial",
  "china",
  "focused",
  "author",
  "self",
  "moving",
  "sell",
  "security",
  "characters",
  "bold",
  "job",
  "joe",
  "whatever",
  "company",
  "james",
  "send",
  "numbers",
  "feature",
  "sent",
  "reality",
  "download",
  "longer",
  "library",
  "mail",
  "main",
  "issues",
  "html",
  "store",
  "giving",
  "story",
  "him",
  "thing",
  "posts",
  "hit",
  "archive",
  "topic",
  "must",
  "performance",
  "posted",
  "both",
  "order",
  "along",
  "recent",
  "haven",
  "wait",
  "special",
  "available",
  "later",
  "third",
  "above",
  "remember",
  "latest",
  "fan",
  "far",
  "probably",
  "sounds",
  "certainly",
  "customers",
  "itself",
  "param",
  "putting",
  "maybe",
  "hot",
  "likely",
  "growth",
  "various",
  "sales",
  "rate",
  "political",
  "widget",
  "between",
  "logo",
  "safe",
  "version",
  "called",
  "tools",
  "david",
  "amazing",
  "design",
  "wasn",
  "comment",
  "team",
  "fix",
  "kids",
  "movie",
  "funny",
  "unique",
  "gone",
  "pick",
  "tech",
  "traffic",
  "house",
  "center",
  "name",
  "nothing",
  "share",
  "entry",
  "groups",
  "sale",
  "large",
  "such",
  "million",
  "ways",
  "couple",
  "fox",
  "comes",
  "beautiful",
  "developer",
  "turned",
  "brain",
  "though",
  "directly",
  "phdo",
  "allow",
  "kind",
  "paper",
  "days",
  "lost",
  "end",
  "hundreds",
  "discussion",
  "understand",
  "personal",
  "stories",
  "charge",
  "did",
  "exactly",
  "reach",
  "fun",
  "radio",
  "republican",
  "launched",
  "original",
  "div",
  "number",
  "width",
  "yes",
  "come",
  "yet",
  "management",
  "showing",
  "weight",
  "save",
  "plans",
  "report",
  "cool",
  "nearly",
  "bad",
  "interview",
  "etc",
  "quickly",
  "worked",
  "bill",
  "weekly",
  "hope",
  "internet",
  "recently",
  "test",
  "fear",
  "support",
  "cost",
  "themselves",
  "gmail",
  "willing",
  "update",
  "missing",
  "war",
  "enough",
  "engine",
  "industry",
  "interface",
  "feed",
  "announced",
  "america",
  "feel",
  "cnn",
  "firefox",
  "trying",
  "web",
  "big",
  "tracker",
  "beyond",
  "development",
  "never",
  "bit",
  "add",
  "issue",
  "chicago",
  "road",
  "type",
  "ads",
  "thought",
  "interested",
  "released",
  "search",
  "computers",
  "powerful",
  "important",
  "releases",
  "government",
  "potential",
  "before",
  "rock",
  "couldn",
  "shockwave",
  "drop",
  "win",
  "drive",
  "telling",
  "higher",
  "others",
  "particularly",
  "box",
  "chance",
  "twitter",
  "buying",
  "player",
  "bottom",
  "taking",
  "alt",
  "learned",
  "choose",
  "padding",
  "things",
  "talks",
  "won",
  "service",
  "access",
  "assets",
  "watching",
  "vice",
  "change",
  "gallery",
  "iphone",
  "japanese",
  "buy",
  "official",
  "app",
  "palin",
  "piece",
  "article",
  "view",
  "further",
  "growing",
  "purchase",
  "sites",
  "away",
  "room",
  "barack",
  "example",
  "art",
  "program",
  "amount",
  "addition",
  "notes",
  "general",
  "voting",
  "usually",
  "everything",
  "point",
  "application",
  "according",
  "given",
  "using",
  "gives",
  "word",
  "face",
  "done",
  "san",
  "whose",
  "saw",
  "say",
  "business",
  "fact",
  "focus",
  "speed",
  "ability",
  "give",
  "info",
  "money",
  "list",
  "california",
  "talking",
  "average",
  "please",
  "michael",
  "morning",
  "popular",
  "competition",
  "beta",
  "set",
  "leaving",
  "none",
  "account",
  "serious",
  "written",
  "steve",
  "matters",
  "close",
  "turn",
  "lead",
  "individual",
  "school",
  "getting",
  "online",
  "six",
  "magazine",
  "billion",
  "page",
  "images",
  "soon",
  "front",
  "event",
  "choice",
  "yesterday",
  "apart",
  "events",
  "address",
  "paid",
  "continue",
  "note",
  "asking",
  "shows",
  "thread",
  "try",
  "final",
  "awesome",
  "trust",
  "might",
  "allowfullscreen",
  "actually",
  "obvious",
  "found",
  "problem",
  "fast",
  "learning",
  "science",
  "mccain",
  "forums",
  "edition",
  "offering",
  "rich",
  "certain",
  "conversation",
  "full",
  "facebook",
  "information",
  "happens",
  "famous",
  "federal",
  "offers",
  "title",
  "person",
  "pdf",
  "machines",
  "language",
  "following",
  "saturday",
  "swf",
  "ideas",
  "project",
  "businesses",
  "past",
  "again",
  "wide",
  "anything",
  "mean",
  "jobs",
  "seeing",
  "interesting",
  "space",
  "already",
  "ends",
  "paul",
  "market",
  "easily",
  "asked",
  "gets",
  "behavior",
  "consider",
  "american",
  "display",
  "php",
  "less",
  "wanted",
  "oct",
  "voters",
  "meet",
  "device",
  "didn",
  "testing",
  "click",
  "photos",
  "start",
  "since",
  "true",
  "state",
  "although",
  "files",
  "john",
  "thousands",
  "users",
  "feedflare",
  "someone",
  "rights",
  "august",
  "sound",
  "held",
  "comments",
  "free",
  "conference",
  "lives",
  "png",
  "perhaps",
  "help",
  "reports",
  "archives",
  "enjoy",
  "evidence",
  "pre",
  "microsoft",
  "risk",
  "pro",
  "needed",
  "mac",
  "mad",
  "power",
  "south",
  "man",
  "map",
  "used",
  "group",
  "told",
  "may",
  "become",
  "daily",
  "user",
  "uses",
  "website",
  "sign",
  "opportunity",
  "bush",
  "aa",
  "ab",
  "ac",
  "ad",
  "ae",
  "af",
  "cover",
  "touch",
  "al",
  "am",
  "unfortunately",
  "ap",
  "works",
  "networks",
  "includes",
  "men",
  "projects",
  "browser",
  "ca",
  "cb",
  "cc",
  "read",
  "cd",
  "ce",
  "cf",
  "forum",
  "vspace",
  "real",
  "image",
  "politics",
  "co",
  "forward",
  "wouldn",
  "mostly",
  "ea",
  "eb",
  "ec",
  "often",
  "ed",
  "ee",
  "ef",
  "yourself",
  "tough",
  "em",
  "en",
  "apple",
  "base",
  "earlier",
  "youtube",
  "named",
  "corporate",
  "register",
  "japan",
  "names",
  "coming",
  "id",
  "im",
  "parents",
  "worst",
  "yahoo",
  "machine",
  "include",
  "websites",
  "increasingly",
  "gave",
  "email",
  "clear",
  "night",
  "current",
  "learn",
  "university",
  "bring",
  "button",
  "mm",
  "mp",
  "apps",
  "mr",
  "margin",
  "candidate",
  "guess",
  "return",
  "target",
  "came",
  "oh",
  "source",
  "key",
  "ol",
  "strike",
  "videos",
  "blogging",
  "phones",
  "os",
  "currently",
  "major",
  "leave",
  "format",
  "half",
  "location",
  "national",
  "qm",
  "always",
  "newspaper",
  "heard",
  "november",
  "hand",
  "known",
  "knows",
  "times",
  "flickr",
  "mdash",
  "chief",
  "heart",
  "until",
  "myself",
  "went",
  "sports",
  "st",
  "windows",
  "former",
  "font",
  "happen",
  "results",
  "case",
  "cash",
  "uk",
  "ul",
  "um",
  "scene",
  "media",
  "rest",
  "huge",
  "seem",
  "seen",
  "hard",
  "doesn",
  "poll",
  "model",
  "creating",
  "obama",
  "form",
  "listen",
  "entire",
  "hate",
  "successful",
  "systems",
  "ym",
  "decision",
  "screen",
  "guys",
  "allows",
  "death",
  "selling",
  "digital",
  "attack",
  "cause",
  "itunes",
  "better",
  "contact",
  "connection",
  "weekend",
  "figure",
  "blank",
  "useful",
  "commercial",
  "race"]

Up to file-list 03/test.rb:

@@ -15,7 +15,9 @@ require 'clusters.rb'
 cs = Clusters.new
 blognames,words,data = cs.readline('myblogdata.txt')
 clust = cs.hcluster(data)
 pp clust

commit 32:	3fc1fa966059
parent 31:	315c41b2a9c9
branch:	default