O'Reilly logo
live online training icon Live Online training

Time Series Analysis for Business Analytics

enter image description here

Working with business data in Python and pandas

Topic: Data
Walter Paczkowski, Ph.D.

Time-oriented data is more common in business applications than you might think. Transactions data, for example, is collected and time-stamped for tracking purposes; production processes are monitored in time for delays, downtime, and material shortages; human resources data is collected daily to track absenteeism, sickness patterns, and insurance claims...the list goes on. And to analyze all this data, you need to apply time series methods and procedures. Yet many business data analysts are unfamiliar with how to work with time data.

Join expert Walter R. Paczkowski for a deep dive into the issues, methods, and tools for working with time-oriented data in Python and pandas. In just three hours, you’ll learn how to work with, extract, convert, aggregate, and visualize time series data and understand the basics of forecast development.

What you'll learn-and how you can apply it

By the end of this live online course, you’ll understand:

  • The importance of time series data in business applications
  • The data cube paradigm
  • Methods for handling and analyzing times series data
  • The basics of times series modeling and forecasting

And you’ll be able to:

  • Create time series indexes
  • Extract time components from datetime identifiers
  • Resample and aggregate time series data
  • Develop a simple but fundamental times series model
  • Develop an h-step-ahead forecast

This training course is for you because...

  • You’re a business data analyst in the private sector.
  • You work with time series data.
  • You want to better understand the capabilities of Python and pandas.

Prerequisites

  • A working knowledge of Python, pandas, and the Jupyter Notebook
  • A basic understanding of statistics, including descriptive statistics, visualization, and OLS models

Recommended preperation:

Recommended follow-up:

About your instructor

  • Walter R. Paczkowski is a market research consultant at Data Analytics Corp., helping companies in a wide range of industries, such as telecommunications, pharmaceuticals, jewelry, food and beverages, and automotive, to mention a few, turn their market data into actionable information. Walter is also an adjunct faculty member of the Department of Economics at Rutgers University. He brings a wealth of knowledge to share about data analysis, drawing on his over 40 years of extensive quantitative experience as an analyst in AT&T's Analytical Support Center, a member of the technical staff at AT&T Bell Labs, head of pricing research in AT&T's Computer Systems Division, and founder of Data Analytics Corp. He was also an adjunct faculty member of the Department of Mathematics and Statistics at the College of New Jersey. Walter is the author of two analytical books—Market Data Analysis Using JMP (SAS Press, 2016) and Pricing Analytics (Routledge, 2018)—with a third forthcoming on quantitative methods for new product development (Routledge, 2020). He holds a PhD in economics from Texas A&M University.

Schedule

The timeframes are only estimates and may vary according to how the class is progressing

Introduction (10 minutes)

  • Presentation: The importance of times series data; defining times series and terminology; issues with time series data; the data cube paradigm—panel data, the DataFrame MultiIndex
  • Q&A

Highlights of pandas' time functionality (30 minutes)

  • Presentation: Datetime variable; date epoch; calendrical functions; DatetimeIndex and PeriodIndex
  • Hands-on exercises: Create a datetime variable; create a DatetimeIndex; create a PeriodIndex
  • Q&A

Time series structure (30 minutes)

  • Presentation: Simple statistics; time series shifts and trends—lags and percent changes; data visualization
  • Hands-on exercises: Create time series statistics; create a time series graph; create a month one lag variable; create a percent change variable; graph the two new variables
  • Q&A

Break (5 minutes)

Digging into time series data with pandas (35 minutes)

  • Presentation: Collapsing the data cube; the accessor function; resampling and aggregating time series
  • Hands-on exercises: Use an accessor to get the year value; resample a DataFrame; aggregate time series data; create a boxplot graph of aggregated data; create a decade variable with a graph
  • Q&A

Break (5 minutes)

Time series modeling: Introduction (60 minutes)

  • Presentation: Econometric models—autocorrelation and Durbin-Watson test; overview of time series models—ARIMA(p, d, q); the AR(1) model; introduction to stationarity; train-test split for time series; model estimation; forecast h-steps ahead
  • Hands-on exercises: Estimate a model; create time series graphs for autocorrelation; check the Durbin-Watson statistic; correct the estimated model for autocorrelation; check for stationarity; do a Dickey-Fuller test; do a first difference correction; estimate an AR(1) model; develop a one-step-ahead forecast; graph the actual and forecasted series
  • Q&A

Wrap-up and Q&A (5 minutes)