def test_init(self):
df = self.df.copy()
powerForecaster = PowerForecaster(df)
self.assertEqual(powerForecaster.df.shape, (37920, 3))
df = self.df.copy()
powerForecaster = PowerForecaster(df, upsample_freq='D')
self.assertEqual(powerForecaster.df.shape, (395, 3))
def test_fit_predict(self):
powerForecaster = PowerForecaster(self.df, Models.LSTM)
powerForecaster.fit()
result = powerForecaster.predict()
if self.interactive:
plt.plot(result)
self.assertTrue(True)
plt.show()
def test_find_index(self):
powerForecaster = PowerForecaster(self.df)
#import pdb; pdb.set_trace()
i, j = find_index(powerForecaster.df, '2012-12-01', '2013-01-01')
self.assertEqual(i, 2880)
self.assertEqual(j, 5855)
j, _ = find_index(powerForecaster.df, "2013-01-01")
self.assertEqual(j, 5856)
def test_plot_duration(self):
powerForecaster = PowerForecaster(self.df, upsample_freq='8H')
df_list = [[
powerForecaster.df[[ColumnNames.LABEL.value]],
powerForecaster.df[[ColumnNames.TEMPERATURE.value]]
]]
plot_data_frames(df_list,
start_date_st="2013-5-01",
end_date_st="2013-11-01")
def test_adjust_index_and_training_shift(self):
powerForecaster = PowerForecaster(self.df.copy(), upsample_freq='8H')
powerForecaster.adjust_index_and_training_shift(
training_duration_in_frequency=30 * 3 # 30 days, freq is 8H
,
start_date_in_labeling_st="2012-12-10",
start_date_training_st="2012-12-01")
# Another call
powerForecaster = PowerForecaster(self.df.copy(), upsample_freq='8H')
powerForecaster.adjust_index_and_training_shift(
training_duration_in_frequency=30 * 3 # 30 days, freq is 8H
,
start_date_in_labeling_st="2012-12-10")
def test_lstm(self):
df = self.df.copy()
powerForecaster = PowerForecaster(df,
model=Models.LSTM,
upsample_freq='8H')
powerForecaster.block_after_date("2013-06-01")
powerForecaster.adjust_index_and_training_shift(
start_date_in_labeling_st="2012-11-02")
powerForecaster.sliding_window()
powerForecaster.fit()
if self.interactive:
powerForecaster.plot_history()
powerForecaster.evaluate()
model = powerForecaster.model_type.value
df_predicted = powerForecaster.lstm_predict(
model,
start_date_to_predict_st="2013-6-01",
duration_in_freq=3 * 30)
def test_block_date(self):
df = self.df.copy()
powerForecaster = PowerForecaster(df,
model=Models.LSTM,
upsample_freq='8H')
powerForecaster.block_after_date("2013-10-01")
def test_sliding_window(self):
powerForecaster = PowerForecaster(self.df, Models.LSTM)
powerForecaster.sliding_window()
window_size = Constants.SLIDING_WINDOW_SIZE_OR_TIME_STEPS
self.assertEqual(powerForecaster.shuffled_X.shape,
(37920 - window_size, window_size))
def test_visual_inspection(self):
powerForecaster = PowerForecaster(self.df)
powerForecaster.visual_inspection()
self.assertTrue(True)
def test_fit_VAR(self):
df = self.df.copy()
powerForecaster = PowerForecaster(df, model=Models.VAR)
powerForecaster.fit()
predicted = powerForecaster.predict()