demo.py

'''
This file runs through a brief demo of how to create a branch, save and load a tree, replicate a subtree, modify
different branches of the tree, execute the tree, and manipulate the results.
Author: Sam Peters
Group name: The Classy Coders
Most recent modification: 2/9/21
'''

import ts_analysis_support as tsas

filename = "data/Time Series Data/1_temperature_test.csv"

def main():
    t = tsas.tree.TransformationTree(tsas.operatorkeys.operator_input_keys, tsas.operatorkeys.operator_output_keys)
    # Creating a single branch
    read = t.add_operator(tsas.preprocessing.TimeSeries.read_from_file, [filename], t.root, tag="read")
    assign_time = t.add_operator(tsas.preprocessing.TimeSeries.assign_time, ["01/01/2000 12:30", 24], read)
    denoise = t.add_operator(tsas.preprocessing.TimeSeries.denoise, [], assign_time)
    ts2db = t.add_operator(tsas.preprocessing.TimeSeries.ts2db, [None, 0.8, 0.1, 0.1, 2, 1, None], denoise, tag="db",
                           save_result=False)
    extract = t.add_operator(tsas.preprocessing.TimeSeries.clip, ["01/01/2020", "01/01/2021"], ts2db)
    model = t.add_operator(tsas.modelingAndForecasting.mlp_model, [], ts2db, tag="model")
    fit = t.add_operator(tsas.modelingAndForecasting.fit, [], model)
    predict = t.add_operator(tsas.modelingAndForecasting.predict, [], fit)
    mse = t.add_operator(tsas.visualization.mse, [], predict, tag="mse", save_result=True)
    plot = t.add_operator(tsas.visualization.plot, [], mse)
    his = t.add_operator(tsas.visualization.histogram, [], plot)  # histogram
    box = t.add_operator(tsas.visualization.box_plot, [], his)  # box plot

    # Save then loading tree, and finding the ts2db operator in the branch we previously created
    tsas.tree.save(t, "demo_tree.sav")
    t = tsas.tree.load("demo_tree.sav")
    ts2db_operator = t.get_nodes_by_tag("db")[0]

    # Replicating the part of the branch at or below the ts2db operator
    db_operators = [t.replicate_subtree(ts2db_operator, tag_modifier=f"_{2 + 1 + i}_features") for i in range(10)]

    # Looping through the new subtrees and modifying their ts2db operators to take in different number of datapoints as input
    counter = 1
    for node in db_operators:
        t.replace_operator(tsas.preprocessing.TimeSeries.ts2db, [None, 0.8, 0.1, 0.1, 2 + counter, 1, None], node)
        counter += 1

    # Executing tree
    t.execute_tree()

    sorted_results = sorted(t.results, key=lambda x: x[0])
    best_branch = sorted_results[0]
    print(f"Best MSE: {best_branch[0]}, from branch: {t.get_path_str(best_branch[1])}")
    

main()