class RankSVM:
"""Python interface to rank svm in svm_light."""
def __init__(self, **kwargs):
self.model = None
def train(self, features, grades):
"""Train a rank_svm on the specified features and grades.
train_rank_svm(self, features, grades):
features - numpy array/matrix with one row per essay
grades - vector with one entry per essay
"""
self.min_grade = min(grades)
self.max_grade = max(grades)
num_essays, num_features = features.shape
# Convert data into svmlight format [(label, [(feature, value), ...], query_id), ...]
training_data = []
for essay_ind,grade in enumerate(grades):
feature_list = [(feat_ind+1,feat_val) for feat_ind,feat_val in enumerate(features[essay_ind,:])]
training_data.append((grade, feature_list, 1))
self.model = svmlight.learn(training_data, type='ranking', verbosity=0, C=100)
grade_counts = {}
for grade in grades:
if grade not in grade_counts:
grade_counts[grade] = 0
grade_counts[grade] += 1
self.grade_probs = dict([(grade, count/float(num_essays)) for grade,count in grade_counts.iteritems()])
scores = self.classify_rank_svm(features)
self.curve = Curve(scores, probs=self.grade_probs)
def grade(self, features, options={}):
scores = self.classify_rank_svm(features)
return [self.curve.curve(score) for score in scores]
def classify_rank_svm(self, features):
"""Run rank_svm to rank the specified essay features (numpy matrix/array).
Returns a vector of scores of the specified essays."""
assert self.model is not None
# Convert data into svmlight format [(label, [(feature, value), ...], query_id), ...]
test_data = []
for essay_ind,feat_vec in enumerate(features):
feature_list = [(feat_ind+1,feat_val) for feat_ind,feat_val in enumerate(feat_vec)]
test_data.append((0, feature_list, 1))
return svmlight.classify(self.model, test_data)
class SVM:
"""Python interface to rank SVM c code."""
tmp_path = "/tmp"
data_file = os.path.join(tmp_path, "svm_rank_features.dat")
model_file = os.path.join(tmp_path, "svm_rank_model.dat")
test_file = os.path.join(tmp_path, "svm_rank_test.dat")
predictions_file = os.path.join(tmp_path, "svm_rank_predictions")
os.system("cd learn && make > /dev/null")
def __init__(self, **kwargs):
pass
def train(self, features, grades):
"""Train a rank_svm on the specified features and grades.
train_rank_svm(self, features, grades):
features - numpy array/matrix with one row per essay
grades - vector with one entry per essay
"""
self.min_grade = min(grades)
self.max_grade = max(grades)
num_essays, num_features = features.shape
f = open(SVM.data_file, "w")
f.write("%d\t%d\n" % (num_essays, num_features))
for essay_ind in range(num_essays):
feature_str = "\t".join([str(feat) for feat in features[essay_ind, :]])
f.write("%d\t%s\n" % (grades[essay_ind], feature_str))
f.close()
os.system("learn/rank_svm %s %s > /dev/null" % (SVM.data_file, SVM.model_file))
# Set a curve based on the SVM ranking scores
grade_counts = {}
for grade in grades:
if grade not in grade_counts:
grade_counts[grade] = 0
grade_counts[grade] += 1
self.grade_probs = dict([(grade, count / float(num_essays)) for grade, count in grade_counts.iteritems()])
scores = self.classify_rank_svm(features)
self.curve = Curve(scores, probs=self.grade_probs)
def grade(self, features, options={}):
scores = self.classify_rank_svm(features)
return [self.curve.curve(score) for score in scores]
def classify_rank_svm(self, features):
"""Run rank_svm to rank the specified essay features (numpy matrix/array).
Returns a vector of scores of the specified essays."""
num_essays, num_features = features.shape
f = open(SVM.test_file, "w")
f.write("%d\t%d\n" % (num_essays, num_features))
for essay_ind in range(num_essays):
feature_str = "\t".join([str(feat) for feat in features[essay_ind, :]])
f.write("0\t%s\n" % feature_str)
f.close()
os.system("learn/rank_svm %s %s %s > /dev/null" % (SVM.test_file, SVM.model_file, SVM.predictions_file))
f = open(SVM.predictions_file)
scores = [float(score) for score in f.readlines()]
f.close()
return scores
