kperry2215

def decompose_time_series(series, desired_frequency):
    """
    Perform STL decomposition on the time series.
    Arguments:
        series: Pandas series. Time series sequence that we wish to decompose.
        desired_frequency: Integer. Time frequency of the series. If we want to detect 
        a yearly trend, we'd set the value equal to 365.
    Outputs:
        Plot of time series STL decomposition.
    """

kperry2215

def isolation_forest_anomaly_detection(df, 
                                       column_name, 
                                       outliers_fraction):
    """
    In this definition, time series anomalies are detected using an Isolation Forest algorithm.
    Arguments:
        df: Pandas dataframe
        column_name: string. Name of the column that we want to detect anomalies in
        outliers_fraction: float. Percentage of outliers allowed in the sequence.
    Outputs:

kperry2215

def low_pass_filter_anomaly_detection(df, 
                                      column_name, 
                                      number_of_stdevs_away_from_mean):
    """
    Implement a low-pass filter to detect anomalies in a time series, and save the filter outputs 
    (True/False) to a new column in the dataframe.
    Arguments:
        df: Pandas dataframe
        column_name: string. Name of the column that we want to detect anomalies in
        number_of_stdevs_away_from_mean: float. Number of standard deviations away from

kperry2215

#Add in a couple anomalous data points for detection by the algorithm
anomaly_dictionary={80: 3.1, 
                    200: 3,
                    333: 1,
                    600: 2.6,
                    710: 2.1,
                    890: 2.3,
                    1100: 1,
                    1211: 2.6,
                    1309: 2.3}

kperry2215

import pandas as pd
import matplotlib.pyplot as plt
import eia

def retrieve_time_series(api, series_ID):
    """
    Return the time series dataframe, based on API and unique Series ID
    Arguments:
        api: API that we're connected to
        series_ID: string. Name of the series that we want to pull from the EIA API

kperry2215

#Create EIA API using your specific API key
api_key = 'YOUR API KEY HERE'
api = eia.API(api_key)
    
#Pull the oil WTI price data
series_ID='PET.RWTC.D'
#Retrieve Data By Series ID 
series_search = api.data_by_series(series=series_ID)
##Create a pandas dataframe from the retrieved time series
price_df = pd.DataFrame(series_search)

kperry2215

def retrieve_time_series(api, series_ID):
    """
    Return the time series dataframe, based on API and unique Series ID
    api: API that we're connected to
    series_ID: string. Name of the series that we want to pull from the EIA API
    """
    #Retrieve Data By Series ID 
    series_search = api.data_by_series(series=series_ID)
    ##Create a pandas dataframe from the retrieved time series
    df = pd.DataFrame(series_search)

kperry2215

def retrieve_time_series(api, series_ID):
    """
    Return the time series dataframe, based on API and unique Series ID
    api: API that we're connected to
    series_ID: string. Name of the series that we want to pull from the EIA API
    """
    #Retrieve Data By Series ID 
    series_search = api.data_by_series(series=series_ID)
    ##Create a pandas dataframe from the retrieved time series
    df = pd.DataFrame(series_search)

kperry2215

#Obtain feature importances in the model
feature_importances = pd.DataFrame(rf.feature_importances_,
                                   index = feature_list,
                                   columns=['importance']).sort_values('importance', 
                                   ascending=False)
print(feature_importances)

kperry2215

#Plug in optimized model parameters into final RF model 
rf = RandomForestClassifier(n_estimators=1100, 
                                max_depth=100,
                                random_state = 1500)
#Fit the model 
rf.fit(train_features, train_labels)
# Use the forest's predict method on the test data
print(confusion_matrix(test_labels, 
                           rf.predict(test_features),
                           labels=['Hour 8', 'Hour 9', 'Hour 10',