def setUp(self):
self.data_location = 'data/daily-min-temperatures.csv'
self.plot_save_location = '/home/boats/Desktop/'
self.target_df = importDataFrame(self.data_location)
self.split_date = pd.to_datetime('1986-01-01 00:00:00')
self.train_df = makeTrainDF(self.target_df, self.split_date)
self.valid_df = makeValidationDF(self.target_df, self.split_date)
self.empty_pred_df = makeEmptyPredictionDF(self.target_df,
self.split_date)
self.pred_df = LastValueModel.create_prediction(
self.train_df, self.empty_pred_df)
self.test_model = PredictionModel('test name')
def dummy_pred_fn(train_df, empty_pred_df):
last_value = (train_df[train_df['dt'] == train_df['dt']
.max()]['val'])
pred_df = empty_pred_df.copy(deep=True)
pred_df['val'] = last_value
return pred_df
self.test_model.predict = dummy_pred_fn
self.pred_df_from_model = self.test_model.predict(self.train_df,
self.empty_pred_df)
self.plot_df = self.test_model.present_results(
self.train_df, self.valid_df, self.pred_df, self.plot_save_location
)
def test_user_journey(self):
# Ivan just downloaded some data and wants to play with it
# He makes a dataframe from the csv
self.df = importDataFrame(self.DATA_LOCATION)
self.assertIsNotNone(self.df)
# He determines the split date and creates train and validation dfs
self.split_date = '1987-01-01'
self.train = makeTrainDF(self.df, self.split_date)
self.valid = makeValidationDF(self.df, self.split_date)
self.empty_pred = makeEmptyPredictionDF(self.df, self.split_date)
# He decides to apply the Naive prediction model
self.LV_pred = LastValueModel.create_prediction(
self.train, self.empty_pred)
# He then wants to admire the results of his hard work in a png file
LastValueModel.present_results(self.train, self.valid, self.LV_pred,
self.PLOT_SAVE_LOCATION)
# He also wants to see the performance of the model on MAE and RMSE
self.MAE_val = MAE.calculate_performance(self.valid, self.LV_pred)
self.RMSE_val = RMSE.calculate_performance(self.valid, self.LV_pred)
self.assertGreater(self.MAE_val, 0.0)
self.assertGreater(self.RMSE_val, 0.0)
# Satisfied, he now moves on to Prophet.
# He creates the prediction
self.PH_pred = ProphetModel.create_prediction(self.train,
self.empty_pred)
# He plots the result in a png file
ProphetModel.present_results(self.train, self.valid, self.PH_pred,
self.PLOT_SAVE_LOCATION)
def setUp(self):
self.data_location = 'data/daily-min-temperatures.csv'
self.plot_name = 'plots/plot.png'
self.target_df = importDataFrame(self.data_location)
self.split_date = pd.to_datetime('1987-01-01 00:00:00')
self.train_df = makeTrainDF(self.target_df, self.split_date)
self.valid_df = makeValidationDF(self.target_df, self.split_date)
self.empty_pred_df = makeEmptyPredictionDF(self.target_df,
self.split_date)
self.ph_pred = ProphetModel.create_prediction(
self.train_df, self.empty_pred_df)
def setUp(self):
self.data_location = 'data/daily-min-temperatures.csv'
self.target_df = importDataFrame(self.data_location)
self.split_date = pd.to_datetime('1987-01-01 00:00:00')
self.train_df = makeTrainDF(self.target_df, self.split_date)
self.valid_df = makeValidationDF(self.target_df, self.split_date)
self.empty_pred_df = makeEmptyPredictionDF(self.target_df,
self.split_date)
self.pred_df = LastValueModel.create_prediction(
self.train_df, self.empty_pred_df)
self.MAPE = LossFunction(name='MAPE')
def calc_MAPE(valid_df, pred_df):
test = valid_df['val'].sub(pred_df['val'])
test = test.sum()
return test
self.MAPE.calculate_performance = calc_MAPE
def setUp(self):
self.data_location = 'data/daily-min-temperatures.csv'
self.plot_name = 'plots/plot.png'
self.target_df = importDataFrame(self.data_location)
self.split_date = pd.to_datetime('1987-01-01 00:00:00')
self.train_df = makeTrainDF(self.target_df, self.split_date)
self.valid_df = makeValidationDF(self.target_df, self.split_date)
self.empty_pred_df = makeEmptyPredictionDF(self.target_df,
self.split_date)
self.ph_pred = ProphetModel.create_prediction(
self.train_df, self.empty_pred_df)
self.train_df['label'] = 'train'
self.valid_df['label'] = 'validation'
self.ph_pred['label'] = 'prediction'
self.merged_df = pd.concat([self.train_df,
self.ph_pred,
self.valid_df])
self.merged_pivot = self.merged_df.pivot_table(index='dt',
columns='label',
values='val')
def setUp(self):
self.data_location = 'data/daily-min-temperatures.csv'
self.target_df = importDataFrame(self.data_location)
self.split_date = pd.to_datetime('1986-01-01 00:00:00')
self.empty_pred_df = makeEmptyPredictionDF(self.target_df,
self.split_date)
def setUp(self):
self.data_location = 'data/daily-min-temperatures.csv'
self.my_test_ts = importDataFrame(self.data_location)
self.column_names = self.my_test_ts.columns