Time Series Forecasting in Python

inside front cover
Time Series Forecasting in Python
Copyright
dedication
contents
front matter
preface
acknowledgments
about this book
Who should read this book?
How this book is organized: A roadmap
About the code
liveBook discussion forum
Author online
about the author
about the cover illustration
Part 1. Time waits for no one
1 Understanding time series forecasting
1.1 Introducing time series
1.1.1 Components of a time series
1.2 Bird’s-eye view of time series forecasting
1.2.1 Setting a goal
1.2.2 Determining what must be forecast to achieve your goal
1.2.3 Setting the horizon of the forecast
1.2.4 Gathering the data
1.2.5 Developing a forecasting model
1.2.6 Deploying to production
1.2.7 Monitoring
1.2.8 Collecting new data
1.3 How time series forecasting is different from other regression tasks
1.3.1 Time series have an order
1.3.2 Time series sometimes do not have features
1.4 Next steps
Summary
2 A naive prediction of the future
2.1 Defining a baseline model
2.2 Forecasting the historical mean
2.2.1 Setup for baseline implementations
2.2.2 Implementing the historical mean baseline
2.3 Forecasting last year’s mean
2.4 Predicting using the last known value
2.5 Implementing the naive seasonal forecast
2.6 Next steps
Summary
3 Going on a random walk
3.1 The random walk process
3.1.1 Simulating a random walk process
3.2 Identifying a random walk
3.2.1 Stationarity
3.2.2 Testing for stationarity
3.2.3 The autocorrelation function
3.2.4 Putting it all together
3.2.5 Is GOOGL a random walk?
3.3 Forecasting a random walk
3.3.1 Forecasting on a long horizon
3.3.2 Forecasting the next timestep
3.4 Next steps
3.5 Exercises
3.5.1 Simulate and forecast a random walk
3.5.2 Forecast the daily closing price of GOOGL
3.5.3 Forecast the daily closing price of a stock of your choice
Summary
Part 2. Forecasting with statistical models
4 Modeling a moving average process
4.1 Defining a moving average process
4.1.1 Identifying the order of a moving average process
4.2 Forecasting a moving average process
4.3 Next steps
4.4 Exercises
4.4.1 Simulate an MA(2) process and make forecasts
4.4.2 Simulate an MA(q) process and make forecasts
Summary
5 Modeling an autoregressive process
5.1 Predicting the average weekly foot traffic in a retail store
5.2 Defining the autoregressive process
5.3 Finding the order of a stationary autoregressive process
5.3.1 The partial autocorrelation function (PACF)
5.4 Forecasting an autoregressive process
5.5 Next steps
5.6 Exercises
5.6.1 Simulate an AR(2) process and make forecasts
5.6.2 Simulate an AR(p) process and make forecasts
Summary
6 Modeling complex time series
6.1 Forecasting bandwidth usage for data centers
6.2 Examining the autoregressive moving average process
6.3 Identifying a stationary ARMA process
6.4 Devising a general modeling procedure
6.4.1 Understanding the Akaike information criterion (AIC)
6.4.2 Selecting a model using the AIC
6.4.3 Understanding residual analysis
6.4.4 Performing residual analysis
6.5 Applying the general modeling procedure
6.6 Forecasting bandwidth usage
6.7 Next steps
6.8 Exercises
6.8.1 Make predictions on the simulated ARMA(1,1) process
6.8.2 Simulate an ARMA(2,2) process and make forecasts
Summary
7 Forecasting non-stationary time series
7.1 Defining the autoregressive integrated moving average model
7.2 Modifying the general modeling procedure to account for non-stationary series
7.3 Forecasting a non-stationary times series
7.4 Next steps
7.5 Exercises
7.5.1 Apply the ARIMA(p,d,q) model on the datasets from chapters 4, 5, and 6
Summary
8 Accounting for seasonality
8.1 Examining the SARIMA(p,d,q)(P,D,Q)m model
8.2 Identifying seasonal patterns in a time series
8.3 Forecasting the number of monthly air passengers
8.3.1 Forecasting with an ARIMA(p,d,q) model
8.3.2 Forecasting with a SARIMA(p,d,q)(P,D,Q)m model
8.3.3 Comparing the performance of each forecasting method
8.4 Next steps
8.5 Exercises
8.5.1 Apply the SARIMA(p,d,q)(P,D,Q)m model on the Johnson & Johnson dataset
Summary
9 Adding external variables to our model
9.1 Examining the SARIMAX model
9.1.1 Exploring the exogenous variables of the US macroeconomics dataset
9.1.2 Caveat for using SARIMAX
9.2 Forecasting the real GDP using the SARIMAX model
9.3 Next steps
9.4 Exercises
9.4.1 Use all exogenous variables in a SARIMAX model to predict the real GDP
Summary
10 Forecasting multiple time series
10.1 Examining the VAR model
10.2 Designing a modeling procedure for the VAR(p) model
10.2.1 Exploring the Granger causality test
10.3 Forecasting real disposable income and real consumption
10.4 Next steps
10.5 Exercises
10.5.1 Use a VARMA model to predict realdpi and realcons
10.5.2 Use a VARMAX model to predict realdpi and realcons
Summary
11 Capstone: Forecasting the number of antidiabetic drug prescriptions in Australia
11.1 Importing the required libraries and loading the data
11.2 Visualizing the series and its components
11.3 Modeling the data
11.3.1 Performing model selection
11.3.2 Conducting residual analysis
11.4 Forecasting and evaluating the model’s performance
Next steps
Part 3. Large-scale forecasting with deep learning
12 Introducing deep learning for time series forecasting
12.1 When to use deep learning for time series forecasting
12.2 Exploring the different types of deep learning models
12.3 Getting ready to apply deep learning for forecasting
12.3.1 Performing data exploration
12.3.2 Feature engineering and data splitting
12.4 Next steps
12.5 Exercise
Summary
13 Data windowing and creating baselines for deep learning
13.1 Creating windows of data
13.1.1 Exploring how deep learning models are trained for time series forecasting
13.1.2 Implementing the DataWindow class
13.2 Applying baseline models
13.2.1 Single-step baseline model
13.2.2 Multi-step baseline models
13.2.3 Multi-output baseline model
13.3 Next steps
13.4 Exercises
Summary
14 Baby steps with deep learning
14.1 Implementing a linear model
14.1.1 Implementing a single-step linear model
14.1.2 Implementing a multi-step linear model
14.1.3 Implementing a multi-output linear model
14.2 Implementing a deep neural network
14.2.1 Implementing a deep neural network as a single-step model
14.2.2 Implementing a deep neural network as a multi-step model
14.2.3 Implementing a deep neural network as a multi-output model
14.3 Next steps
14.4 Exercises
Summary
15 Remembering the past with LSTM
15.1 Exploring the recurrent neural network (RNN)
15.2 Examining the LSTM architecture
15.2.1 The forget gate
15.2.2 The input gate
15.2.3 The output gate
15.3 Implementing the LSTM architecture
15.3.1 Implementing an LSTM as a single-step model
15.3.2 Implementing an LSTM as a multi-step model
15.3.3 Implementing an LSTM as a multi-output model
15.4 Next steps
15.5 Exercises
Summary
16 Filtering a time series with CNN
16.1 Examining the convolutional neural network (CNN)
16.2 Implementing a CNN
16.2.1 Implementing a CNN as a single-step model
16.2.2 Implementing a CNN as a multi-step model
16.2.3 Implementing a CNN as a multi-output model
16.3 Next steps
16.4 Exercises
Summary
17 Using predictions to make more predictions
17.1 Examining the ARLSTM architecture
17.2 Building an autoregressive LSTM model
17.3 Next steps
17.4 Exercises
Summary
18 Capstone: Forecasting the electric power consumption of a household
18.1 Understanding the capstone project
18.1.1 Objective of this capstone project
18.2 Data wrangling and preprocessing
18.2.1 Dealing with missing data
18.2.2 Data conversion
18.2.3 Data resampling
18.3 Feature engineering
18.3.1 Removing unnecessary columns
18.3.2 Identifying the seasonal period
18.3.3 Splitting and scaling the data
18.4 Preparing for modeling with deep learning
18.4.1 Initial setup
18.4.2 Defining the DataWindow class
18.4.3 Utility function to train our models
18.5 Modeling with deep learning
18.5.1 Baseline models
18.5.2 Linear model
18.5.3 Deep neural network
18.5.4 Long short-term memory (LSTM) model
18.5.5 Convolutional neural network (CNN)
18.5.6 Combining a CNN with an LSTM
18.5.7 The autoregressive LSTM model
18.5.8 Selecting the best model
18.6 Next steps
Part 4. Automating forecasting at scale
19 Automating time series forecasting with Prophet
19.1 Overview of the automated forecasting libraries
19.2 Exploring Prophet
19.3 Basic forecasting with Prophet
19.4 Exploring Prophet’s advanced functionality
19.4.1 Visualization capabilities
19.4.2 Cross-validation and performance metrics
19.4.3 Hyperparameter tuning
19.5 Implementing a robust forecasting process with Prophet
19.5.1 Forecasting project: Predicting the popularity of “chocolate” searches on Google
19.5.2 Experiment: Can SARIMA do better?
19.6 Next steps
19.7 Exercises
19.7.1 Forecast the number of air passengers
19.7.2 Forecast the volume of antidiabetic drug prescriptions
19.7.3 Forecast the popularity of a keyword on Google Trends
Summary
20 Capstone: Forecasting the monthly average retail price of steak in Canada
20.1 Understanding the capstone project
20.1.1 Objective of the capstone project
20.2 Data preprocessing and visualization
20.3 Modeling with Prophet
20.4 Optional: Develop a SARIMA model
20.5 Next steps
21 Going above and beyond
21.1 Summarizing what you’ve learned
21.1.1 Statistical methods for forecasting
21.1.2 Deep learning methods for forecasting
21.1.3 Automating the forecasting process
21.2 What if forecasting does not work?
21.3 Other applications of time series data
21.4 Keep practicing
Appendix. Installation instructions
Installing Anaconda
Python
Jupyter Notebooks
GitHub Repository
Installing Prophet
Installing libraries in Anaconda
index
inside back cover