def test_train_save_load_predict(self):
print "test train, save, load and predict..."
xs = SklearnRandomForestTrainTestModel.get_xs_from_results(
self.features)
ys = SklearnRandomForestTrainTestModel.get_ys_from_results(
self.features)
xys = SklearnRandomForestTrainTestModel.get_xys_from_results(
self.features)
self.model = SklearnRandomForestTrainTestModel(
{
'norm_type': 'normalize',
'random_state': 0
}, None)
self.model.train(xys)
self.model.to_file(self.model_filename)
self.assertTrue(os.path.exists(self.model_filename))
loaded_model = SklearnRandomForestTrainTestModel.from_file(
self.model_filename, None)
result = loaded_model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.17634739353518517, places=4)
def test_train_save_load_predict(self):
print "test train, save, load and predict..."
xs = SklearnRandomForestTrainTestModel.get_xs_from_results(self.features)
ys = SklearnRandomForestTrainTestModel.get_ys_from_results(self.features)
xys = SklearnRandomForestTrainTestModel.get_xys_from_results(self.features)
self.model = SklearnRandomForestTrainTestModel({'norm_type':'normalize', 'random_state':0}, None)
self.model.train(xys)
self.model.to_file(self.model_filename)
self.assertTrue(os.path.exists(self.model_filename))
loaded_model = SklearnRandomForestTrainTestModel.from_file(self.model_filename, None)
result = loaded_model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.17634739353518517, places=4)
def test_train_predict_randomforest(self):
print "test random forest train and predict..."
# random forest don't need proper data normalization
xs = SklearnRandomForestTrainTestModel.get_xs_from_results(
self.features, [0, 1, 2])
ys = SklearnRandomForestTrainTestModel.get_ys_from_results(
self.features, [0, 1, 2])
xys = SklearnRandomForestTrainTestModel.get_xys_from_results(
self.features, [0, 1, 2])
model = SklearnRandomForestTrainTestModel(
{
'norm_type': 'normalize',
'random_state': 0
}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643766, places=4)
model = SklearnRandomForestTrainTestModel(
{
'norm_type': 'clip_0to1',
'random_state': 0
}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)
model = SklearnRandomForestTrainTestModel(
{
'norm_type': 'custom_clip_0to1',
'custom_clip_0to1_map': {
'Moment_noref_feature_1st_score': [0.0, 100.0],
},
'random_state': 0
}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)
model = SklearnRandomForestTrainTestModel(
{
'norm_type': 'clip_minus1to1',
'random_state': 0
}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)
model = SklearnRandomForestTrainTestModel(
{
'norm_type': 'none',
'random_state': 0
}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)
def test_train_predict_randomforest(self):
print "test random forest train and predict..."
# random forest don't need proper data normalization
xs = SklearnRandomForestTrainTestModel.get_xs_from_results(self.features, [0, 1, 2])
ys = SklearnRandomForestTrainTestModel.get_ys_from_results(self.features, [0, 1, 2])
xys = SklearnRandomForestTrainTestModel.get_xys_from_results(self.features, [0, 1, 2])
model = SklearnRandomForestTrainTestModel({'norm_type':'normalize',
'random_state': 0}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643766, places=4)
model = SklearnRandomForestTrainTestModel({'norm_type':'clip_0to1',
'random_state': 0}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)
model = SklearnRandomForestTrainTestModel(
{'norm_type': 'custom_clip_0to1',
'custom_clip_0to1_map': {
'Moment_noref_feature_1st_score': [0.0, 100.0],
},
'random_state': 0
}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)
model = SklearnRandomForestTrainTestModel({'norm_type':'clip_minus1to1',
'random_state': 0}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)
model = SklearnRandomForestTrainTestModel({'norm_type':'none', 'random_state': 0}, None)
model.train(xys)
result = model.evaluate(xs, ys)
self.assertAlmostEquals(result['RMSE'], 0.051804171170643752, places=4)