def test3_read_from_file_pandas():
#df = pd.read_csv(r'../timeSeriesData/TimeSeriesData1/1_temperature_test.csv')
#df = pd.read_csv('../timeSeriesData/TimeSeriesData2/AtmPres2005NovMin.csv')
df = fio.read_from_file('../timeSeriesData/TimeSeriesData2/AtmPres2005NovMin.csv')
print(df)
#print(df)
#print(d2)
d3 = fio.read_from_file('../timeSeriesData/TimeSeriesData2/AtmPres2005NovMin.csv')
print(df.iloc[:,-1:])
#print(d3.iloc[:,-3])
return 0
def smape(y_forecast, y_test: str): """ takes in a database (y_forcast) and a file name (y_test) and returns the symmetric mean absolute percentage error Author: Nick Titzler """ yf = pre.db2ts(y_forecast) yt = fio.read_from_file(y_test) return 100/len(yf.iloc[:,-1].to_numpy()) * np.sum(2 * np.abs(yt.iloc[:,-1].to_numpy() - yf.iloc[:,-1].to_numpy()) / (np.abs(yf.iloc[:,-1].to_numpy()) + np.abs(yt.iloc[:,-1].to_numpy())))
def test2_timeSeriesData2(): fname1 = "../timeSeriesData/TimeSeriesData2/AtmPres2005NovMin.csv" ts = fio.read_from_file(fname1) fname2 = "../timeSeriesData/TimeSeriesData2/wind_aristeomercado_10m_complete.csv" ts = fio.read_from_file(fname2) fname3 = "../timeSeriesData/TimeSeriesData1/9_distribution_subsampled_train_empty.csv" fname4 = "../timeSeriesData/TimeSeriesData1/AtmPres2005NovMinEmpty.csv" fname5 = "../timeSeriesData/TimeSeriesData1/save1.p" ts1 = fio.read_from_file(fname5) print(ts1) ts = fio.read_from_file(fname4) print(ts)
def mape(y_forecast, y_test: str): """ takes in a database (y_forcast) and a file name (y_test) and returns the mean absoulute percentage error Author: Nick Titzler """ yf = pre.db2ts(y_forecast) yt = fio.read_from_file(y_test) return np.mean((np.abs(yf.iloc[:,-1].to_numpy()-yt.iloc[:,-1].to_numpy()) / yf.iloc[:,-1].to_numpy())) * 100
def mse(y_forecast, y_test: str): """ takes in a database (y_forcast) and a file name (y_test: str) and returns the mean squared error between the datasets Author: Nick Titzler """ yf = pre.db2ts(y_forecast) yt = fio.read_from_file(y_test) return mean_squared_error(yf.iloc[:,-1].to_numpy(), yt.iloc[:,-1].to_numpy())
def mlp_forecast(model_data, x_filename):
"""
Takes in a tuple containing a MLPRegressor object and a tuple as
well a string. Returns a numpy matrix.
Predicts a future set of values from a given set of values
and a trained model.
"""
# extract model and tree from model data
model = model_data[0]
window = model_data[1]
# grab test data from file
x = fio.read_from_file(x_filename)
x = x.to_numpy()
# predict values
y_hat = model.predict(x)
# interpolate predicted values to real size
y_hat = mlp_output_mapper(y_hat, window)
return y_hat
def test1_timeSeriesData1():
"""
test time series creation from timeSeriesData1 files
"""
fileNames = ["1_temperature_test.csv","1_temperature_train.csv","2_temperature_subsampled_test.csv",
"2_temperature_subsampled_train.csv", "3_passengers_test.csv","3_passengers_train.csv","4_irradiance_test.csv",
"4_irradiance_train.csv", "5_irradiance_subsampled_test.csv", "5_irradiance_subsampled_train.csv", "6_sunspots_test.csv", "6_sunspots_train.csv",
"7_distribution_subsampled_norm_test.csv", "7_distribution_subsampled_norm_train.csv", "8_distribution_subsampled_test.csv", "8_distribution_subsampled_train.csv"]
fname = "../timeSeriesData/TimeSeriesData1/1_temperature_test.csv"
try:
for item in fileNames:
fname = "../timeSeriesData/TimeSeriesData1/"+item
ts = fio.read_from_file(fname)
except:
print("error in test 1")
def test_execute_pipeline(test_tree):
tree = TS_Tree()
tree.replace_node("longest_continuous_run", 0)
tree.add_node("impute_missing_data", 0)
tree.add_node("assign_time", 1, data_start=1.0, increment=.2)
tree.add_node("plot", 2)
#tree.add_node("clip", 2, data_start=1.0, data_end=10.0)
print("\n##Test executting a pipeline to node 3##")
tree.print_tree()
fname1 = "../timeSeriesData/TimeSeriesData2/AtmPres2005NovMin.csv"
ts = fio.read_from_file(fname1)
results = tree.execute_path(ts, 3)
tree2 = TS_Tree()
tree2.replace_node(
"read_from_file",
0,
input_filename="../timeSeriesData/TimeSeriesData2/AtmPres2005NovMin.csv"
)
def ts2db(input_file, perc_train, perc_val, perc_test, input_index,
output_index, output_file):
"""
Takes in an input data file to read in the time series, as well as
how much the user wants the data to be split into three categories:
training, validation, and testing. Then, it creates three separate time
series that it turns into machine learning model friendly databases with
the input and output sizes provided, and returns those.
"""
# read in time series data from file
ts = fio.read_from_file(input_file)
# split time series data into training, validation, and test sets
ts_splits = split_data(ts, perc_train, perc_val, perc_test)
# convert datasets into databases that can be processed by
# a machine learning model
train_db = design_matrix(ts_splits[0], input_index,
output_index) # CHANGE OUTPUT HANDLING
val_db = design_matrix(ts_splits[1], input_index, output_index)
test_db = design_matrix(ts_splits[2], input_index, output_index)
# return set of databases
return (train_db, val_db, test_db)
def test1_write_to_file():
ts = TS.TimeSeries()
ts = fio.read_from_file("../timeSeriesData/TimeSeriesData1/1_temperature_test.csv")
fio.write_to_file(ts, "1_temperature_test_output.csv")
def test4_empty_file():
df = fio.read_from_file('../timeSeriesData/TimeSeriesData1/oneItem.csv')
print(df)