Understanding time series data

One of the fundamental things when dealing with any dataset is to get intimate with it: without understanding what you are dealing with, you cannot build a successful statistical model.

Getting ready

To execute this recipe, you will need pandas, Statsmodels, and Matplotlib. No other prerequisites are required.

How to do it…

One of the fundamental statistics to check for any time series is the autocorrelation function (ACF), partial autocorrelation function (PACF), and spectral density (the ts_timeSeriesFunctions.py file):

import statsmodels as sm

# read the data
riverFlows = pd.read_csv(data_folder + 'combined_flow.csv', 
    index_col=0, parse_dates=[0])

# autocorrelation function
acf = {}    # to store the results

for col in riverFlows.columns:
    acf[col] = sm.tsa.stattools.acf(riverFlows[col])

# partial autocorrelation function
pacf = {}

for col in riverFlows.columns:
    pacf[col] = sm.tsa.stattools.pacf(riverFlows[col])

# periodogram (spectral density...