def step06(paramFile):
#util = ParameterUtil(parameter_file = 'data/formatado/arxiv/nowell_example_1994_1999.txt')
util = ParameterUtil(parameter_file=paramFile)
myparams = Parameterization(util.keyword_decay, util.lengthVertex, util.t0,
util.t0_, util.t1, util.t1_,
util.FeaturesChoiced, util.graph_file,
util.trainnig_graph_file, util.test_graph_file,
util.decay)
calc = Calculate(myparams, util.nodes_notlinked_file, util.calculated_file,
util.ordered_file, util.maxmincalculated_file)
topRank = Analyse.getTopRank(util.analysed_file + '.random.analised.txt')
calc.Ordering_separating_File(topRank)
weightedScoresChoiced=util.WeightedScoresChoiced,
FullGraph=None)
myparams.generating_Training_Graph()
myparams.generating_Test_Graph()
selection = VariableSelection(myparams.trainnigGraph,
util.nodes_notlinked_file, util.min_edges)
calc = Calculate(myparams, util.nodes_notlinked_file, util.calculated_file,
util.ordered_file, util.maxmincalculated_file)
calc.Separating_calculateFile()
analise = Analyse(
myparams, FormatingDataSets.get_abs_file_path(util.calculated_file),
FormatingDataSets.get_abs_file_path(util.analysed_file) +
'.random.analised.txt', calc.qtyDataCalculated)
topRank = Analyse.getTopRank(util.analysed_file + '.random.analised.txt')
calc.Ordering_separating_File(topRank)
for OrderingFilePath in calc.getfilePathOrdered_separeted():
analise = Analyse(myparams, OrderingFilePath,
OrderingFilePath + '.analised.txt', topRank)
print "Trainning Period:", myparams.t0, " - ", myparams.t0_
print "Test Period:", myparams.t1, " - ", myparams.t1_
print "# Papers in Trainning: ", myparams.get_edges(myparams.trainnigGraph)
print "# Authors in Training: ", myparams.get_nodes(myparams.trainnigGraph)
print "# Papers in Test: ", myparams.get_edges(myparams.testGraph)
print "# Authors in Test", myparams.get_nodes(myparams.testGraph)
print "# pair of Authors with at least 3 articles Calculated: ", calc.qtyDataCalculated #FormatingDataSets.getTotalLineNumbers(FormatingDataSets.get_abs_file_path(util.calculated_file))
print "# pair of Authors that is connected in Test Graph in a random way: ", topRank
print "Max values found in calculations: ", str(calc.maxValueCalculated)
