def exportWeights(self, _features, _file):
M = CSV()
if len(self.classes) > 0:
M.header = ['class0', 'class1'] + _features
else:
M.header = _features
for c in range(len(self.weights)):
W = self.weights[c]
F = []
for feature in _features:
if feature in W:
F.append(W[feature])
else:
F.append(0)
if len(self.classes) > 0:
M.data.append(','.join(self.classes[c] + [str(x) for x in F]))
else:
M.data.append(','.join([str(x) for x in F]))
M.save(_file)
def run(self, _training, _model, _batchSize, _resultFile): csv = CSV(_training) csv.randomize(1000) csv.removeIndices() R = ResultMatrix() for i in range(int(len(csv.data)/_batchSize)): c = CSV() c.header = csv.header c.data = csv.data[0:(i+1)*_batchSize] file = self.resultFolder + "subset_" + str(i) + ".csv" c.save(file) header, data = Experiment(file).regression([_model], 10) R.add(header, data) R.save(_resultFile)
# perform a 10-fold cross validation
e = Experiment(training, "example_feature_reduction")
e.regression([model], 10)
CSV(e.path("cv_0.csv")).save(e.path("subset_0.csv"))
xTicks = ["None"]
# obtain a feature ranking
M = CSV(e.path("features_0.csv")).toMatrix()
M.normalizeRows()
M.sortByMean()
# sequentially remove the least important feature from the training data and retrain the model
subset = e.path("subset.csv")
for i in range(len(M.header) - 1):
key = M.header[-1]
M.header = M.header[0:-1]
csv.removeColumnWithKey(key)
csv.save(subset)
e = Experiment(subset, "example_feature_reduction")
e.regression([model], 10)
CSV(e.path("cv_0.csv")).save(e.path("subset_" + str(i + 1) + ".csv"))
xTicks.append(key)
#
files = [e.path("subset_" + str(i) + ".csv") for i in range(len(xTicks))]
ResultVisualizer().boxplots(files,
"r2",
xTicks,
xlabel='Sequentially Removed Features',
ylabel='R2',