Introduction to Time Series Forecasting with Python - How to Prepare Data and Develop Models to Predict the Future

Copyright
Welcome
I Fundamentals
Python Environment
Why Python?
Python Libraries for Time Series
Python Ecosystem Installation
Summary
What is Time Series Forecasting?
Time Series
Time Series Nomenclature
Describing vs. Predicting
Components of Time Series
Concerns of Forecasting
Examples of Time Series Forecasting
Summary
Time Series as Supervised Learning
Supervised Machine Learning
Sliding Window
Sliding Window With Multivariates
Sliding Window With Multiple Steps
Summary
II Data Preparation
Load and Explore Time Series Data
Daily Female Births Dataset
Load Time Series Data
Exploring Time Series Data
Summary
Basic Feature Engineering
Feature Engineering for Time Series
Goal of Feature Engineering
Minimum Daily Temperatures Dataset
Date Time Features
Lag Features
Rolling Window Statistics
Expanding Window Statistics
Summary
Data Visualization
Time Series Visualization
Minimum Daily Temperatures Dataset
Line Plot
Histogram and Density Plots
Box and Whisker Plots by Interval
Heat Maps
Lag Scatter Plots
Autocorrelation Plots
Summary
Resampling and Interpolation
Resampling
Shampoo Sales Dataset
Upsampling Data
Downsampling Data
Summary
Power Transforms
Airline Passengers Dataset
Square Root Transform
Log Transform
Box-Cox Transform
Summary
Moving Average Smoothing
Moving Average Smoothing
Data Expectations
Daily Female Births Dataset
Moving Average as Data Preparation
Moving Average as Feature Engineering
Moving Average as Prediction
Summary
III Temporal Structure
A Gentle Introduction to White Noise
What is a White Noise?
Why Does it Matter?
Is your Time Series White Noise?
Example of White Noise Time Series
Summary
A Gentle Introduction to the Random Walk
Random Series
Random Walk
Random Walk and Autocorrelation
Random Walk and Stationarity
Predicting a Random Walk
Is Your Time Series a Random Walk?
Summary
Decompose Time Series Data
Time Series Components
Combining Time Series Components
Decomposition as a Tool
Automatic Time Series Decomposition
Summary
Use and Remove Trends
Trends in Time Series
Shampoo Sales Dataset
Detrend by Differencing
Detrend by Model Fitting
Summary
Use and Remove Seasonality
Seasonality in Time Series
Minimum Daily Temperatures Dataset
Seasonal Adjustment with Differencing
Seasonal Adjustment with Modeling
Summary
Stationarity in Time Series Data
Stationary Time Series
Non-Stationary Time Series
Types of Stationary Time Series
Stationary Time Series and Forecasting
Checks for Stationarity
Summary Statistics
Augmented Dickey-Fuller test
Summary
IV Evaluate Models
Backtest Forecast Models
Model Evaluation
Monthly Sunspots Dataset
Train-Test Split
Multiple Train-Test Splits
Walk Forward Validation
Summary
Forecasting Performance Measures
Forecast Error (or Residual Forecast Error)
Mean Forecast Error (or Forecast Bias)
Mean Absolute Error
Mean Squared Error
Root Mean Squared Error
Summary
Persistence Model for Forecasting
Forecast Performance Baseline
Persistence Algorithm
Shampoo Sales Dataset
Persistence Algorithm Steps
Summary
Visualize Residual Forecast Errors
Residual Forecast Errors
Daily Female Births Dataset
Persistence Forecast Model
Residual Line Plot
Residual Summary Statistics
Residual Histogram and Density Plots
Residual Q-Q Plot
Residual Autocorrelation Plot
Summary
Reframe Time Series Forecasting Problems
Benefits of Reframing Your Problem
Minimum Daily Temperatures Dataset
Naive Time Series Forecast
Regression Framings
Classification Framings
Time Horizon Framings
Summary
V Forecast Models
A Gentle Introduction to the Box-Jenkins Method
Autoregressive Integrated Moving Average Model
Box-Jenkins Method
Identification
Estimation
Diagnostic Checking
Summary
Autoregression Models for Forecasting
Autoregression
Autocorrelation
Minimum Daily Temperatures Dataset
Quick Check for Autocorrelation
Autocorrelation Plots
Persistence Model
Autoregression Model
Summary
Moving Average Models for Forecasting
Model of Residual Errors
Daily Female Births Dataset
Persistence Forecast Model
Autoregression of Residual Error
Correct Predictions with a Model of Residuals
Summary
ARIMA Model for Forecasting
Autoregressive Integrated Moving Average Model
Shampoo Sales Dataset
ARIMA with Python
Rolling Forecast ARIMA Model
Summary
Autocorrelation and Partial Autocorrelation
Minimum Daily Temperatures Dataset
Correlation and Autocorrelation
Partial Autocorrelation Function
Intuition for ACF and PACF Plots
Summary
Grid Search ARIMA Model Hyperparameters
Grid Searching Method
Evaluate ARIMA Model
Iterate ARIMA Parameters
Shampoo Sales Case Study
Daily Female Births Case Study
Extensions
Summary
Save Models and Make Predictions
Process for Making a Prediction
Daily Female Births Dataset
Select Time Series Forecast Model
Finalize and Save Time Series Forecast Model
Make a Time Series Forecast
Update Forecast Model
Extensions
Summary
Forecast Confidence Intervals
ARIMA Forecast
Daily Female Births Dataset
Forecast Confidence Interval
Interpreting the Confidence Interval
Plotting the Confidence Interval
Summary
VI Projects
Time Series Forecast Projects
5-Step Forecasting Task
Iterative Forecast Development Process
Suggestions and Tips
Summary
Project: Monthly Armed Robberies in Boston
Overview
Problem Description
Test Harness
Persistence
Data Analysis
ARIMA Models
Model Validation
Extensions
Summary
Project: Annual Water Usage in Baltimore
Overview
Problem Description
Test Harness
Persistence
Data Analysis
ARIMA Models
Model Validation
Summary
Project: Monthly Sales of French Champagne
Overview
Problem Description
Test Harness
Persistence
Data Analysis
ARIMA Models
Model Validation
Summary
VII Conclusions
How Far You Have Come
Further Reading
Applied Time Series
Time Series
Machine Learning
Getting Help
Contact the Author
VIII Appendix
Standard Time Series Datasets
Shampoo Sales Dataset
Minimum Daily Temperatures Dataset
Monthly Sunspots Dataset
Daily Female Births Dataset
Airline Passengers Dataset
Workaround for Saving ARIMA Models
Daily Female Births Dataset
Python Environment
ARIMA Model Save Bug
ARIMA Model Save Bug Workaround
Summary