/sample-outline

A course outline for ARIMA forecasting in Python

Primary LanguageJupyter Notebook

Chapter 1 - Introduction to timeseries forecasting

  • Lesson 1.1 - Introduction to timeseries
    • A learning objective: Using decomposition to understand the primary components of a timeseries: seasonality, trend, and error.
  • Lesson 1.2 - Autocorrelation and differencing
    • A learning objective: Generate ACF/PCF plots, extract meaning from them, relate to differencing methods
  • Lesson 1.3 - Stationarity and white noise
    • A learning objective: Recognize timeseries that are stationary, understand the relation to forecasting. Introduce the random walk model(I).

Chapter 2 - Choosing parameters for ARIMA

  • Lesson 2.1 - Introduction to Autoregressive models(AR)
    • A learning objective: Define AR models; provide examples; show indicators for AR parameter value.
  • Lesson 2.2 - Introduction to Moving Average models(MA)
    • A learning objective: Define MA models; provide examples; show indicators for MA parameter value.
  • Lesson 2.3 - Introduction to ARIMA models(AR + I + MA)
    • A learning objective: Define ARIMA models; provide examples; show indicators for parameter values.
  • Lesson 2.4 - Practice modeling with ARIMA
    • A learning objective: Take multiple timeseries and use previous techniques to choose an effective model.

Chapter 3 - Forecasting, Performance, and Confidence

  • Lesson 3.1 - Using ARIMA to forecast
    • A learning objective: Generate some forecasts using models previously fit.
  • Lesson 3.2 - Evaluating forecasts
    • A learning objective: Conduct In/Out of sample validation, compare metrics for forecast performance
  • Lesson 3.3 - Confidence
    • A learning objective: Generate confidence intervals for forecasts, understand their meaning.
  • Lesson 3.4 - Selecting the best model by validation
    • A learning objective: Choose between reasonable models by comparing performance and criterion.

Chapter 4 - Seasonal ARIMA with Exogenous

  • Lesson 4.1 - Identifying seasonal components
    • A learning objective: Define SARIMA models; provide examples; show indicators for parameter values.
  • Lesson 4.2 - Building SARIMA models
    • A learning objective: Compare ARIMA and SARIMA models; identify parameters.
  • Lesson 4.3 - Exogenous variables
    • A learning objective: Use dummy variables to add correlation effects to your model.
  • Lesson 4.4 - Forecasting with SARIMAX
    • A learning objective: Identify parameters for a timeseries; show the progression of performance from the addition of model components; produce forecasts.