Autoplay
Autocomplete
Previous Lesson
Complete and Continue
Feature Engineering for Time Series Forecasting
Welcome
Introduction (7:39)
Course curriculum (9:44)
Course overview (6:54)
Course requirements (1:53)
Course material (1:40)
Download Jupyter notebooks
Download presentations
Download datasets
Refer a friend program
Tabularizing time series data
Time series forecasting (9:52)
Forecasting models (6:20)
Datasets, features and targets (6:55)
Forecasting framework (6:58)
Feature engineering overview (9:32)
Quiz: tabularizing time series data
Forecasting demo: data analysis (10:12)
Forecasting demo: feature engineering (14:19)
Forecasting demo: training the forecaster (6:21)
Code assignment - tabularize time series (3:46)
Summary (8:23)
Challenges in feature engineering for forecasting
Challenges in feature engineering (4:26)
Machine learning workflow (4:02)
Feature engineering in tabular data (9:42)
Feature engineering in forecasting - considerations (10:05)
Feature engineering in forecasting - pipelines (4:43)
Quiz: machine learning workflow
Forecasting demo - intro (2:22)
Feature engineering pipeline - demo (9:24)
Forecasting one step ahead: demo (6:50)
Code assignment - feature engineering pipeline (3:00)
Multistep forecasting (4:02)
Direct forecasting (5:28)
Direct forecasting: demo (9:58)
Recursive forecasting (5:41)
Recursive forecasting: demo (11:47)
Recursive forecasting: multiple horizons - demo (7:12)
Summary (3:32)
Extra Treat: Our Reading Suggestion 📕
Time Series Decomposition
Components of a time series (7:10)
White noise (9:01)
Additive and multiplicative models (7:13)
Log transform (5:27)
Box-Cox transform (11:51)
Box-Cox transform: Guerrero method (12:07)
Box-Cox transform: demo (part 1) (8:50)
Box-Cox transform: demo (part 2) (5:26)
Moving average (13:59)
Moving averages in Pandas: demo (7:03)
Classical decomposition: trend (8:27)
Classical decomposition: seasonality (9:58)
Classical decomposition: demo (5:48)
LOWESS: Theory (11:26)
LOWESS: Practice (6:42)
LOWESS to extract trend: demo (12:29)
LOWESS vs LOESS (6:33)
STL Overview (13:01)
STL theory part 1: LOESS and cycle-subseries (8:43)
STL theory part 2: the inner loop (13:21)
STL theory part 3: the outer loop (4:30)
STL to compute seasonality and trend: demo (7:37)
Multi-seasonal time series (6:40)
Multi-seasonal decomposition (5:57)
MSTL (13:12)
MSTL: demo (11:28)
Summary (14:09)
Added Treat: A Movie We Recommend 🍿
Missing Data Imputation
Imputation overview (5:58)
Forward and backward filling (3:03)
Forward and backward filling: demo (5:58)
Linear interpolation (4:36)
Linear interpolation: demo (5:51)
Spline interpolation (4:51)
Spline interpolation: demo (3:43)
Seasonal decomposition and interpolation (2:32)
Seasonal decomposition and interpolation: demo (6:42)
Summary (6:04)
Outliers
Outliers overview (9:38)
Outliers in time series (5:21)
Rolling statistics (8:27)
Rolling mean for outlier detection (8:31)
Rolling mean for outlier detection: demo (10:13)
Rolling median for outlier detection (7:11)
Rolling median for outlier detection: demo (9:16)
Residuals for outlier detection (8:39)
LOWESS for outlier detection (5:03)
LOWESS and residuals for outlier detection: demo (10:07)
STL for outlier detection (3:58)
STL and residuals for outlier detection: demo (10:02)
Dummy variables to handle outliers and special events (9:58)
Summary (8:44)
More Wisdom: Our Chosen Podcast Episode 🎧
Lag Features
Lag features (8:41)
Lag features: demo (5:04)
How to choose the lags (11:17)
Autoregressive (AR) processes (15:02)
Lag plots (12:54)
Lag plots: demo (6:32)
Autocorrelation function (part 1) (10:58)
Autocorrelation function (part 2) (13:51)
Autocorrelation function: demo (8:57)
Partial autocorrelation function (part 1) (10:20)
Partial autocorrelation function (part 2) (13:59)
Partial autocorrelation function: demo (13:30)
Cross correlation function (part 1) (5:49)
Cross correlation function (part 2) (14:28)
Cross correlation function: demo (14:26)
Distributed lag features (10:24)
Creating good lag features demo: air pollution dataset (5:35)
Creating good lag features demo: domain knowledge (13:50)
Creating good lag features demo: feature selection & modelling (11:53)
Creating good lag features demo: correlation methods (part 1) (11:20)
Creating good lag features demo: correlation methods (part 2) (10:04)
Summary (10:00)
Window Features
Window features overview (2:34)
Rolling window features: definition (4:50)
Rolling window features: picking the window size and statistics (8:37)
Rolling window features: implementation in Python (9:17)
Rolling window features: demo (12:21)
Expanding window features: definition (2:30)
Expanding window features: use cases (3:45)
Expanding window features: implementation in Python (3:41)
Expanding window features: demo (4:48)
Weighted window functions: definition & use cases (11:08)
Weighted window functions: implementation in Python (5:28)
Weighted window functions: demo (11:49)
Exponential weights: definition (5:21)
Exponential weights: expanding windows and implementation (6:17)
Exponential weights: demo (12:31)
Selecting window features: demo (7:43)
Summary (9:47)
Trend Features
Trend features: overview (2:44)
Types of trend (7:07)
Linear trend: using time as a feature (12:39)
Time feature: creating the feature demo (9:07)
Time feature: forecasting demo (13:48)
Non-linear trend: using time as a feature (10:30)
Non-linear time features: creating the features demo (4:52)
Non-linear time features: forecasting demo (6:27)
Recursive forecasting with lags, windows, and trend (8:10)
Trend features and recursive forecasting: demo (14:32)
Piecewise regression and changepoints (part 1) (9:10)
Piecewise regression and changepoints (part 2) (8:22)
Changepoint features: creating the features demo (5:31)
Changepoint features: forecasting demo (9:03)
Tree-based models and trend (8:54)
Tree-based models and trend: detrending with sktime demo (13:21)
Tree-based models and trend: forecasting demo (7:08)
Linear trees using LightGBM (9:05)
Linear trees using LightGBM: demo (6:32)
Summary (6:07)
Seasonality Features
Seasonality and cyclical patterns overview (5:24)
Seasonal lag features (6:38)
Seasonal lag features: demo (11:26)
Date and time features for seasonality (3:58)
Date and time features: demo (part 1) (7:05)
Date and time features: demo (part 2) (6:51)
Why linear models struggle with date and time features (4:09)
Seasonal dummy features (5:49)
Seasonal dummy features: demo (part 1) (5:35)
Seasonal dummy features: demo (part 2) (5:50)
Fourier features: theory (7:41)
Fourier features: how to create and use Fourier features (10:22)
Fourier features: demo (part 1) (10:52)
Fourier features: demo (part 2) (4:24)
Summary (3:33)
Date and Time Features
Date and time features (4:20)
Date features: demo (17:18)
Time features: demo (4:14)
Datetime features with Feature-engine (5:36)
Periodic or Cyclical Features (6:29)
Periodic Features: demo (8:23)
Calendar holidays (2:26)
Calendar holidays: demo (12:14)
Summary (1:05)
Categorical Features
Categorical features - intro (3:13)
One hot encoding (6:03)
One hot encoding with open source (3:58)
One hot encoding: demo (7:51)
Ordinal encoding (1:49)
Ordinal encoding with open source (1:45)
Ordinal encoding: demo (3:34)
Mean encoding (8:53)
Mean encoding: demo with Feature-engine (3:18)
Mean encoding: demo with expanding windows (5:03)
Summary (5:34)
Final section | Next steps
Congratulations
Next steps
Calendar holidays: demo
Lesson content locked
If you're already enrolled,
you'll need to login
.
Enroll in Course to Unlock