svmlab 1.0.0

data/lib/svmprediction.rb

@@ -0,0 +1,176 @@
+require 'arraymethods.rb'
+
+class SVMPrediction < Hash
+  
+  # --- difficultFeature ---
+  def difficultFeature(features)
+    if !features.is_a? SVMFeature
+      raise "Please give a SVMFeature object as argument."
+    end
+    keys = self.keys
+    names = (1...features.dim).map do |i|
+      features.featname(i)
+    end
+    maxlen = names.map{|n| n.size}.max
+    correlations = (1...features.dim).map do |i|
+      predmiss = keys.map { |k|
+        (self[k]['truth'] - self[k]['pred']).abs }
+      feat = keys.map { |k| features[k][i] }
+      "#{names[i-1].rjust(maxlen)}   %.2f"%correlation(predmiss, feat)
+    end
+    correlations.sort_by{|line| line.split.last.to_f.abs}.reverse
+  end
+
+  # --- [] ---
+  # If indexing with a regular expression, a new SVMPrediction object is created
+  # containing all elements with matching keys.
+  def [](expr)
+    if expr.is_a? Regexp
+      subs = SVMPrediction.new
+      self.find_all { |(k,v)| k =~ expr }.each do |i|
+        subs[i[0]] = i[1]
+      end
+      subs
+    else
+      super(expr)
+    end
+  end
+
+  # --- (minus) ---
+  # Removes all predictions in arg that is found in self.
+  def -(arg)
+    self.inject(SVMPrediction.new) do |subs, (k,v)|
+      if !arg[k]
+        subs[k] = v if !arg[k]
+      end
+      subs
+    end
+  end
+
+  # Each should be an array giving more than one plot
+  def genericplot(plotdata, file, title='Plot', xtitle='X', ytitle='Y')
+    Gnuplot.open do |gp| # This could be either a file or the gnuplot process that we pipe to
+      Gnuplot::Plot.new( gp ) do |plot|
+        plot.title  title
+        plot.xlabel xtitle
+        plot.ylabel ytitle
+        plot.set "grid"
+        if file =~ /(png)|(ps)$/
+          plot.terminal "png size 1024,800 font '/usr/share/fonts/truetype/ttf-bitstream-vera/VeraBd.ttf' 20" if file =~ /png$/
+          plot.terminal "postscript eps color" if file =~ /ps$/          
+          plot.output file
+        end
+        plot.data = plotdata
+      end
+    end
+    nil
+  end
+  
+  # --- plot ---
+  # Plots true value on the X axis vs. predicted value on the Y axis.
+  def plot(legends = [], title = 'SVM Prediction', err = nil, file = '')
+    # For historical reasons
+    predarr = [ self ]
+    # Set up dataarr
+    dataarr = predarr.map do |predictions|
+      x, y = predictions.inject([[],[]]) { |data,(example,val)|
+        data[0].push(val['truth'])
+        data[1].push(val['pred'])
+        data }
+    end
+    
+    from = dataarr.inject(dataarr[0][0][0]) { |m,a|
+      [m, a[0].min, a[1].min].min }.floor
+    to = dataarr.inject(dataarr[0][0][0]) { |m,a|
+      [m, a[0].max, a[1].max].max }.ceil
+    sampleindex = 0
+    # Fiddling with legends
+    legends = dataarr.map{|d| "Sample #{sampleindex+=1}"} if legends.size==0
+    err = ['rmsd','cc'] if !err # Default behaviour
+    err = [err] if err.is_a? String
+    if err
+      legends = legends.map { |legend|
+        legend + ' (' +
+        err.map { |e|
+          begin
+            args = if e.split(/,/).size==1 then nil
+                   else '(' + (e.split(/,/)[1..-1]).join(',') + ')' end
+            "#{e} = ".upcase + "%.2f"%eval("self.#{e.split(/,/).first.downcase}#{args}")
+          rescue
+            $!
+          end
+        }.join(', ') + ')'
+      }
+    end
+    # Setting plotdata
+    plotdata = [ Gnuplot::DataSet.new( [[from,to], [from,to]] ) { |ds|
+                   ds.using = '1:2'
+                   ds.with = "lines"
+                   ds.title = "Correct diagonal"
+                   ds.linewidth = 1
+                   ds.matrix = nil } ] +
+      dataarr.zip(legends).inject([]) { |arr,((x,y),legend)|
+        arr.push(Gnuplot::DataSet.new( [x,y] ) { |ds|
+                   ds.using = '1:2'
+                   ds.with = "points"
+                   ds.title = legend
+                   ds.linewidth = 2
+                   ds.matrix = nil }) }
+    genericplot(plotdata, file, title, 'Experimental value', 'Predicted value')
+  end
+
+  # --- rmsd ---
+  # Root mean square deviation
+  def rmsd
+    a = [] # Don't initialize in one line : x=y=[]
+    b = [] # If doing that, they will both refer to the same array
+    self.each { |example,val|
+      a.push(val['truth'].to_f)
+      b.push(val['pred'].to_f)
+    }
+    if (x = a.zip(b).inject(0) {|sd, (d, q)| sd + (d - q)**2 }.to_f / a.length) > 0
+      Math.sqrt(x)
+    else
+      0
+    end
+  end
+
+  # --- cc ---
+  # Correlation coefficient
+  def cc
+    a = [] # Don't initialize in one line : x=y=[]
+    b = [] # If doing that, they will both refer to the same array
+    self.each { |example,val|
+      a.push(val['truth'].to_f)
+      b.push(val['pred'].to_f)
+    }
+    amean = a.inject(0) {|sum,i| sum + i} / a.size.to_f
+    bmean = b.inject(0) {|sum,i| sum + i} / b.size.to_f
+    ssa = a.inject(0.0) {|ss,i| ss + (i-amean)**2}
+    ssb = b.inject(0) {|ss,i| ss + (i-bmean)**2}
+    ssab = a.zip(b).inject(0) {|ss,(ai,bi)| ss + (ai-amean) * (bi-bmean)}
+    if ssab > 0 
+      Math::sqrt(ssab**2 / (ssa * ssb))
+    else
+      0
+    end
+  end
+
+  # --- f1 ---
+  # Precision Recall measure
+  def f1( border = 0)
+    tp = self.select { |k,v| v['truth'] >= border and v['pred'] >= border}.size.to_f
+    fp = self.select { |k,v| v['truth'] < border  and v['pred'] >= border}.size.to_f
+    fn = self.select { |k,v| v['truth'] >= border  and v['pred'] < border}.size.to_f
+    precision = if (denom=tp+fp) > 0 then tp / denom
+                else 0 end
+    recall = if (denom=tp+fn) > 0 then tp / denom
+             else 0 end
+    if (denom = precision + recall) > 0
+      2 * precision * recall / denom
+    else
+      0
+    end
+  end
+
+end

data/lib/test.cfg

@@ -0,0 +1,12 @@
+SVM: 
+  C: 1 #step exp1.0
+  e: 3 #step exp1.0
+  g: 0.4 #step exp0.5
+  Optimization: 
+    Method: patternsearch
+    Nhalf: 3
+    Measure: cc
+  Scale:
+    Default: max
+  LibSVMDataset: /home/fred/SVMLab/trunk/lib/testdata
+  #PosClassFrom: 1.7

data/lib/test.rb

@@ -0,0 +1,5 @@
+require 'wirble'
+Wirble.init
+Wirble.colorize
+
+puts "Welcome!"

data/lib/testdata

@@ -0,0 +1,3 @@
+1 1:2.0 2:6.0
+-1 1:2.0 2:10.0
+-1 1:2.0 2:14.0

data/lib/texput.log

@@ -0,0 +1,20 @@
+This is pdfTeXk, Version 3.141592-1.40.3 (Web2C 7.5.6) (format=latex 2008.11.29)  16 MAR 2009 10:48
+entering extended mode
+ %&-line parsing enabled.
+**main.tex
+
+! Emergency stop.
+<*> main.tex
+            
+*** (job aborted, file error in nonstop mode)
+
+ 
+Here is how much of TeX's memory you used:
+ 3 strings out of 94102
+ 111 string characters out of 1165832
+ 47703 words of memory out of 1500000
+ 3383 multiletter control sequences out of 10000+50000
+ 3640 words of font info for 14 fonts, out of 1200000 for 2000
+ 637 hyphenation exceptions out of 8191
+ 0i,0n,0p,1b,6s stack positions out of 5000i,500n,6000p,200000b,5000s
+No pages of output.

data/lib/tmp.irb.rc

@@ -0,0 +1,81 @@
+require 'irb/completion'
+require 'svmlab'
+
+IRB.conf[:PROMPT][:SVMLab] = {       # name of prompt mode
+  :PROMPT_I => "SVMLab:%03n:%i> ",   # normal prompt
+  :PROMPT_S => "SVMLab:%03n:%i%l ",  # prompt for continuing strings
+  :PROMPT_C => "SVMLab:%03n:%i* ",   # prompt for continuing statement
+  :RETURN => "    ==>%s\n"           # format to return value
+}
+
+IRB.conf[:PROMPT_MODE] = :SVMLab
+
+
+# Sebastian Delmont
+# Pretty print methods
+ANSI_BOLD       = "\033[1m"
+ANSI_RESET      = "\033[0m"
+ANSI_LGRAY    = "\033[0;37m"
+ANSI_GRAY     = "\033[1;30m"
+
+def pm(obj, *options) # Print methods
+  methods = obj.methods
+  methods -= Object.methods unless options.include? :more
+  filter = options.select {|opt| opt.kind_of? Regexp}.first
+  methods = methods.select {|name| name =~ filter} if filter
+
+  data = methods.sort.collect do |name|
+    method = obj.method(name)
+    if method.arity == 0
+      args = "()"
+    elsif method.arity > 0
+      n = method.arity
+      args = "(#{(1..n).collect {|i| "arg#{i}"}.join(", ")})"
+    elsif method.arity < 0
+      n = -method.arity
+      args = "(#{(1..n).collect {|i| "arg#{i}"}.join(", ")}, ...)"
+    end
+    klass = $1 if method.inspect =~ /Method: (.*?)#/
+    [name, args, klass]
+  end
+  max_name = data.collect {|item| item[0].size}.max
+  max_args = data.collect {|item| item[1].size}.max
+  data.each do |item| 
+    print " #{ANSI_BOLD}#{item[0].rjust(max_name)}#{ANSI_RESET}"
+    print "#{ANSI_GRAY}#{item[1].ljust(max_args)}#{ANSI_RESET}"
+    print "   #{ANSI_LGRAY}#{item[2]}#{ANSI_RESET}\n"
+  end
+  data.size
+end
+
+# To enable history saving between sessions
+IRB.conf[:SAVE_HISTORY] = 100
+
+# Stian Haklev / Joel VanderWerf
+# To reduce lengthy output
+class IRB::Context
+  attr_accessor :max_output_size
+
+  alias initialize_before_max_output_size initialize
+  def initialize(*args)
+    initialize_before_max_output_size(*args)
+    @max_output_size = IRB.conf[:MAX_OUTPUT_SIZE] || 80
+  end
+end
+
+class IRB::Irb
+  def output_value
+    text =
+      if @context.inspect?
+        sprintf @context.return_format, @context.last_value.inspect
+      else
+        sprintf @context.return_format, @context.last_value
+      end
+    max = @context.max_output_size
+    if text.size < max
+      puts text
+    else
+      puts text[0..max-1] + "..." + text[-2..-1]
+    end
+  end
+end 

data/lib/v6.cfg

@@ -0,0 +1,124 @@
+--- 
+#------------------------------------------------------------
+# CONFIGURATION OF SVM PARAMETERS
+#------------------------------------------------------------
+SVM: 
+  C: 1 step exp1.0
+  e: 0.3 step exp1.0
+  g: 0.04 step exp1.0
+  #C: 1 #step exp1.0
+  #e: 0.17 #step exp1.0
+  #g: 0.4 #step exp1.0
+  Optimization: 
+    Method: "patternsearch"
+    Nhalf: 3
+  Scale:
+    ddG: 
+    - max #step exp0.1
+    KFCprop:
+    - max #step exp0.1     # size
+    - max #step exp0.1     # chemical properties
+    - max #step exp0.1
+    - max #step exp0.1
+    KFCN:
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    KFCS:
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    - max #step exp0.1
+    KFCPolarBonds:
+    - max
+    KFCGenericBonds:
+    - max
+    KFCHbonds:
+    - max #step exp0.1
+    HBPlusBonds:
+    - max
+    zdscore:
+    - max
+    - max
+    - max
+#------------------------------------------------------------
+# CONFIGURATION OF FEATURE PARAMETERS
+#------------------------------------------------------------
+Feature: 
+  #PosClassFrom: 2.0
+  # Features is a list of all features used
+  # Each feature need to have its configuration given
+  Features: 
+  - ddG
+  - KFCN
+  - KFCS
+  - KFCPolarBonds
+  - KFCGenericBonds
+  - KFCHbonds
+  - KFCprop
+  #- HBPlusBonds
+  #- zdscore
+  # "BaseDir" is the directory from which all paths are derived
+  BaseDir: /home/fred/DeltaDeltaG/Features/
+  # "DataSet" is the file (relative to BaseDir or absolute) 
+  # that holds names of examples
+  DataSet: KFCdatasettest
+  # "Methods" is the .rb file that holds all feature methods
+  Methods:
+    - KFCfeatures.rb
+    - fredrik-hbplus.rb
+    - /home/fred/DeltaDeltaG/ZDOCK-Features/zdock-svm.rb
+  Groups: (0..3)
+  #---SEPARATE FEATURE CONFIGURATIONS FROM HERE--------------
+  #---------------------------ddG----------------------------
+  ddG:
+    Dimensions: 1
+    # "HomeDir" is optional. If not given, it will be set to <BaseDir>/<Feature name>
+    HomeDir: ddG/
+    # "Method" is mandatory and is the name of the method calculating the feature
+    Method: kfc_ddG
+  #----------------------------KFCprop-----------------------
+  KFCprop:
+    Dimensions: 4
+    Method: kfcProperties
+  #----------------------------KFCN--------------------------
+  KFCN:
+    Dimensions: 10
+    Method: kfcN
+  #----------------------------KFCS--------------------------
+  KFCS:
+    Dimensions: 10
+    Method: kfcS
+  #----------------------------KFCPolarBonds-----------------
+  KFCPolarBonds:
+    Method: kfcPolarBonds
+  #----------------------------KFCGenericBonds---------------
+  KFCGenericBonds:
+    Method: kfcGenericBonds
+  #----------------------------KFCHbonds---------------------
+  KFCHbonds:
+    Method: kfcHbonds
+  #----------------------------HBPlusBonds-------------------
+  HBPlusBonds:
+    Dimensions: 1
+    Method: hbonds
+    MaxH2O: 2
+    OnlySideChain: yes
+  #----------------------------zdscore-----------------------
+  zdscore:
+    Dimensions: 3
+    Method: zdscoreSimple
+    HomeDir: /home/fred/DeltaDeltaG/ZDOCK-Features/zdscoreSimple/

metadata

@@ -0,0 +1,102 @@
+--- !ruby/object:Gem::Specification 
+name: svmlab
+version: !ruby/object:Gem::Version 
+  prerelease: false
+  segments: 
+  - 1
+  - 0
+  - 0
+  version: 1.0.0
+platform: ruby
+authors: 
+- Fredrik Johansson
+autorequire: svmlab
+bindir: bin
+cert_chain: []
+
+date: 2010-05-26 00:00:00 +09:00
+default_executable: 
+dependencies: 
+- !ruby/object:Gem::Dependency 
+  name: forkoff
+  prerelease: false
+  requirement: &id001 !ruby/object:Gem::Requirement 
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        segments: 
+        - 0
+        version: "0"
+  type: :runtime
+  version_requirements: *id001
+- !ruby/object:Gem::Dependency 
+  name: gnuplot
+  prerelease: false
+  requirement: &id002 !ruby/object:Gem::Requirement 
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        segments: 
+        - 0
+        version: "0"
+  type: :runtime
+  version_requirements: *id002
+description: 
+email: fredjoha@gmail.com
+executables: []
+
+extensions: []
+
+extra_rdoc_files: []
+
+files: 
+- lib/svmlab-optim.rb
+- lib/v6.cfg
+- lib/tmp.irb.rc
+- lib/libsvmdata.rb
+- lib/testdata
+- lib/test.cfg
+- lib/svmfeature2.rb
+- lib/test.rb
+- lib/svmlab-config.rb
+- lib/svmlab-plot.rb
+- lib/svmlab.rb
+- lib/irb.history
+- lib/arraymethods.rb
+- lib/svmfeature.rb
+- lib/svmlab-irb.rb
+- lib/texput.log
+- lib/svmprediction.rb
+- lib/README
+has_rdoc: true
+homepage: http://ediacara.bmr.kyushu-u.ac.jp/fredrik
+licenses: []
+
+post_install_message: 
+rdoc_options: []
+
+require_paths: 
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      segments: 
+      - 0
+      version: "0"
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      segments: 
+      - 0
+      version: "0"
+requirements: []
+
+rubyforge_project: 
+rubygems_version: 1.3.6
+signing_key: 
+specification_version: 3
+summary: A tool for experimenting with Support Vector Machines.
+test_files: []
+