Time Series Forecasting in Python


Units: 6


Time Series Forecasting is something of a dark horse in the field of data science. It is one of the most applied data science techniques in business - used extensively in finance, in supply chain management, and in production and inventory planning. Moreover, it has a well established theoretical grounding in statistics and dynamic systems theory. Yet, it retains something of an outsider status in data science compared to more recent and popular machine learning methods such as image recognition and natural language processing. Consequently, Time Series Forecasting gets little or no treatment at all in introductory data science and machine learning courses. 

This course is intended to provide a comprehensive introduction to forecasting methods without deep diving into the theoretical details behind each method. Although, the references at the end of each week will fill in many of those details.

The course is intended for the following three audiences.

  1. Graduate students studying in STEM or business fields. 

  2. People doing forecasting in business who may not have had any formal training in the area.

  3. MBA students doing a data elective.

  4. Also relevant for those studying public policy, healthcare management, and related disciplines.

Learning Outcomes

  • Forecaster’s Toolbox - Build and apply time series forecasting models in a variety of business contexts using tools that are useful for many different forecasting situations such as the following.

  • Benchmark/Simple forecasting methods.

  • Ways of making the forecasting task simpler using transformations and adjustments.

  • Residual analysis - methods for checking whether a forecasting method has adequately utilized the available information.

  • Techniques for computing prediction intervals.

  • Be able to use a range of forecasting methods - ETS (Error, Trend, Seasonality), Holt Winter’s, ARIMA (Autoregressive, Integrated, Moving Average), SARIMA, VAR (Vector Autoregression), and a suite of machine learning models such as XGBoost, Random Forest, Support Vector Regression, etc.

  • Be able to use graphical methods to explore the time series data, analyze the validity of the models fitted, and present the forecasting results.

  • Develop skills, mindsets, and behaviors that are most sought after in the industry today - data science, product management, entrepreneurship, and storytelling. 

  • Don’t think like a Data Scientist / Software Engineer / Solutions Architect / Tech Consultant. Think like a CEO. Take an end-to-end view.

Prerequisites Description

  • It is assumed that course participants are familiar with 
    • Introductory statistics.
    • High-school algebra.
    • Graphical methods for describing quantitative data - line plots, histograms, barplots, box plots, correlation plots. 
    • Numerical measures of Central Tendency. 
    • Numerical  measures of Variability.
    • Using the Mean and Standard Deviation to describe data. 
    • Joining data - inner, outer, right, and left join.
    • Hypothesis testing.
    • Simple linear and multiple regression - model assumptions, fitting the model, assessing utility of model.
    • Splitting data into test-train splits and cross validation.
    • Model performance evaluation metrics - RMSE, MAE, MAPE.
  • Basic Python knowledge required - ability to read data and ability to perform basic manipulations using pandas and numpy libraries. 
  • If you don’t have prior experience with the aforementioned, you can still enroll if you have the willingness to learn. 
  • Bring your enthusiasm and curiosity.