XGBoost - Tweaking Parameters 2 - BEST MAE 0.484 (Lag Model) 🚀

DS4B 203-R: High-Performance Time Series Forecasting

Welcome to High-Performance Forecasting

Prerequisites

Prerequisite - R for Business Analysis (DS4B 101-R)

Getting Help

Getting Help (IMPORTANT!!!)

Module 0 - Introduction to High-Performance Forecasting

High-Performance Forecasting - What You're Learning, Why You're Learning It (0:43)

0.1 Forecast Competition Review

0.2 Course Projects - Google Analytics, Email Subscribers, & Sales Forecasting

The Business Case - Developing a Best-in-Class Forecasting System (3:03)

0.3 What Tools are in Your Toolbox?

Module 01 - Time Series Jumpstart

Time Series Jumpstart (0:54)

1.1 Time Series Project Setup

1.2 Business Understanding & Dataset Terminology

1.3 TS Jumpstart: Dive into Forecasting Email Subscribers!

1.3.1 Exploratory Data Analysis for Time Series

1.3.2 Evaluation & Train/Test Windows

1.3.3 Forecasting with Prophet

1.3.4 Forecasting with Feature Engineering

1.3.5 Recap & Code Checkpoint - Module 01 - TS Jumpstart

✨[Part 1] Time Series with Timetk

Welcome to Part 1 - Time Series with Timetk! (2:17)

Module 02 - Time Series Visualization

2.1 Time Series Plots [MUST KNOW FUNCTION] 💡

2.2 Autocorrelation Plots

2.3 Seasonality Plots

2.4 Anomaly Plots

2.5 STL Decomposition & Regression Plots

2.6 Regression Plots [SECRET WEAPON FOR FEATURE ENGINEERING]

2.7 Code Checkpoint - Module 02 - Visualization

Code Checkpoint (File Download)

Module 03 - Time Series Data Wrangling

Setup (File Download) & Overview - Data Wrangling (2:34)

3.1 Summarise By Time [MUST KNOW] 💡

3.2 Pad by Time

3.3 Filter By Time

3.4 Mutate By Time

Extrapolate the Mean, Median, Max, Min By Time (7:57)

3.5 Joining By Time

3.6 Time Series Index Operations

3.7 Forecasting with Future Frames 📈

3.8 Code Checkpoint - Module 03 - Data Wrangling

Code Checkpoint (File Download)

Module 04 - Transformations for Time Series

4.1 Variance Reduction Transformations - Log & Box Cox [MUST KNOW] 💡

4.2 Rolling & Smoothing Transformations

4.3 Range Reduction Transformations

4.4 Imputation & Outlier Cleaning

4.5 Lags & Differencing Transformations [MUST KNOW] 💡

4.6 Fourier Series [MUST KNOW] 💡

4.7 Constrained Interval Forecasting [FORECAST SPOTLIGHT] 📈

4.8 Code Checkpoint - Module 04 - Transformations

Code Checkpoint (File Download)

⛰️ Challenge #1 - Exploring Transactions & Web Page Traffic

Module 05 - Introduction to Feature Engineering (for Time Series)

5.1 Time-Based Features (Trend & Seasonal/Calendar) [MUST KNOW] 💡

5.2 Interactions

Interaction Features - Spikes Every Other Wednesday (7:35)

5.3 Fourier Features

5.4 Autocorrelated Lag Features

5.5 Special Event Features

5.6 External Regressors (Xregs)

5.7 Recommended Model Features

5.8 Code Checkpoint - Module 05 - Introduction to Feature Engineering

Code Checkpoint (File Download)

Module 06 - Advanced Feature Engineering Workflow

6.1 Creating the "Full" Dataset - Extending & Adding Lagged Features & Events [IMPORTANT] 💡

6.2 Separate into Modeling Data & Forecast Data

Data Prepared / Future Data Split (2:48)

6.3 Separate into Training Data & Testing Data

Train / Test Split (3:55)

6.4 Recipes - Feature Engineering Pipeline Steps

6.5 Building the Spline Model

6.6 Introduction to Modeltime Workflow

6.7 Building the Lag Model

6.8 Forecasting the Future

6.9 Saving the Artifacts

6.10 Code Checkpoint - Module 06 - Advanced Feature Engineering

Code Checkpoint (File Download)

⛰️ Challenge #2 - Feature Engineering & Modeltime Workflow [YOU'VE GOT THIS!]

Challenge #2 - Solution

Challenge #2 Bonus - Regularization

Part 1 Complete - You rock! 🙌🙌🙌

WOOO HOOO - You crushed it!

✨[Part 2] Machine Learning for Time Series with Modeltime

Picking Up From Part 1 (Project Download)

Module 07 - Modeltime Workflow [DEEP DIVE] 🌊

7.1 Making Models - Object Types & Requirements

7.2 Modeltime Table

Modeltime Table - Key Requirements (4:27)

7.3 Calibration Table

Calibration Table - How It Works (3:29)

7.4 Measuring Model Accuracy [IMPORTANT!!!]

7.5 Forecasting the Test Data

7.6 Model Refitting & Forecasting

7.7 Code Checkpoint - Module 07A - Modeltime Workflow [In-Depth]

Code Checkpoint (File Download)

7.8 New Features of Modeltime 0.1.0 - Module 07B 🆕

7.9 Code Checkpoint - Module 07B - Modeltime New Features!

Code Checkpoint (File Download)

Module 08 - ARIMA

8.1 ARIMA Concepts 💡

8.2 ARIMA in Modeltime

8.3 Modeltime Auto ARIMA

8.4 Recap - ARIMA

8.5 Code Checkpoint - Module 08 - ARIMA

Code Checkpoint (File Download)

Module 09 - Prophet

9.1 Prophet with Modeltime

9.2 Prophet Concepts 💡

9.3 Back to Modeling with Prophet - XREGS!

9.4 Recap - Prophet

9.5 Checkpoint - Module 09 - Prophet

Code Checkpoint (File Download)

Module 10 - Exponential Smoothing, TBATS, & Seasonal Decomposition

10.1 Exponential Smoothing

10.2 TBATS

10.3 Seasonal Decomposition Models

10.4 Evaluation

10.5 Recap - ETS, TBATS, Seasonal Decomp

10.6 Code Checkpoint - Module 10 - ETS, TBATS, & Seasonal Decomposition

Code Checkpoint (File Download)

⛰️ Challenge #3 - ARIMA + Prophet + ETS + TBATS

Challenge #3 - Solution

Challenge #3 BONUS - ARIMA & Prophet vs Linear Model

11.0 Machine Learning Algorithms [IMPORTANT] 💡

Welcome to Machine Learning for Time Series (File Download) (5:22)

11.1 Elastic Net Algorithm (GLMNet) - Linear

*** Plotting Utility *** - Let's make a helper function to speed evaluation up!

11.2 Multiple Adaptive Regression Splines (MARS) - Linear

11.3 Support Vector Machine (SVM) - Polynomial

11.4 Support Vector Machine (SMV) - Radial Basis Function

11.5 [Important Concept] KNN & Tree-Based Algorithms - The Problem with Predicting Time Series Trends

11.5 K-Nearest Neighbors (KNN) - Similarity (Distance) Based

You're kicking butt... But, don't forget to take breaks

[COFFEE BREAK] With Bill Murray

11.6 Random Forest (Tree-Based)

11.7 XGBoost (Gradient Boosting Machine) - Tree-Based

11.8 Cubist - Combo of Trees (Rules) + Linear Models at Nodes

11.9 Neural Net (NNET) - Like a Linear Regression but Better

11.10 NNETAR - Combining AR Terms with a NNET!

11.11 Modeltime Experimentation Review

11.12 Saving Your Work - Artifacts!

11.13 Checkpoint - Module 11 - Machine Learning Algorithms

Code Checkpoint (File Download)

12.0 Boosted Algorithms - Prophet Boost & ARIMA Boost

Boosted Algorithms - A Powerful Technique for Improving Performance (3:37)

12.1A Prophet Baseline Model

12.1B Prophet Boost

12.2 ARIMA Boost

12.3 Boosted Models - Modeltime Workflow

12.4 Code Checkpoint - Boosted Algorithms

Code Checkpoint (File Download)

13.0 Hyper Parameter Tuning & Cross Validation - For Time Series

13.1 Reviewing 28 Models (It's Easy with Modeltime)

13.2 [SEQUENTIAL MODELS] NNETAR - Hyperparameter Tuning Process

13.2.1 - NNETAR Tuning, Round 1 - Default Params

13.2.2 NNETAR Tuning, Round 2 - Finding the Sweet Spot!

13.3 [NON-SEQUENTIAL MODELS] Prophet Boost - Hyperparameter Tuning Process

13.3.1 Prophet Boost Tuning, Round 1 - Default Parameters

13.3.2 Prophet Boost Tuning, Round 2 - Controlling Learning Rate

13.3.3 Prophet Boost Tuning, Round 3 - Honing In

13.4 Saving Our Progress

Recap & Saving the Models (6:53)

13.5 Code Checkpoint - Model 13 - Hyperparameter Tuning

Code Checkpoint (File Download)

14.0 Ensemble Time Series Models (Stacking)

14.1 Model Performance Review

14.2 Average Ensemble

14.3 Weighted Average Ensembles

14.4.A Stacked Ensembles - Stacking Process

14.4.B Stacked Ensembles - Stacking with Tunable Algorithms

14.5 Multi-Level Stacking

14.6 Modeltime Workflow for Ensembles

14.7 Saving Your Work

14.8 Code Checkpoint - Module 14 - Ensemble Methods

Code Checkpoint [File Download]

15.0 Forecasting at Scale - Time Series Groups [Panel Data]

15.1 Data Understanding & Preparation

15.2 Time Splitting - Train/Test

Time Series Split (Train/Test) (3:50)

15.3 Preprocessing & Recipes

15.4 Modeling: Make 7 Panel Models

15.5 Hyperparameter Tuning the Panel Models

15.6 Modeltime Panel Evaluation

15.7 Time Series Cross Validation (Modeltime Resample)

15.8 Ensemble Panel Models

15.9 Recap

🎉 Congrats - You Just Forecasted 20 Time Series Using Panel Data Techniques! (2:28)

15.10 Code Checkpoint - Panel Data

Code Checkpoint [File Download]

✨ [PART 3] Deep Learning with GluonTS

Module 16 - Setting Up GluonTS & Intro to Reticulate

16.1 Configuring the "r-gluonts" Python Environment

16.2 Testing Modeltime GluonTS 🧪

16.3 Getting to Know Reticulate & Your Python Environments 🐍

16.4 Using a Custom GluonTS Environment

16.5 Code Checkpoint - GluonTS Environment Setup

Code Checkpoint [File Download]

Module 17 - Using GluonTS with Reticulate

17.1 Reticulated Python Basics

17.2 Data, Preprocessing, & GluonTS ListDatasets

17.3 DeepAR GluonTS Model

17.4 Visualizing the Forecast | matplotlib, ggplot, & plotly

17.5 Introducing Modeltime GluonTS

17.6 Saving & Loading GluonTS Models

17.7 🎁Bonus!! GluonTS Deep Factor Models

17.8 Conclusions & Pro/Cons

Reticulated GluonTS vs Modeltime GluonTS (Pros & Cons) (4:43)

17.9 Code Checkpoint - GluonTS Reticulate

Code Checkpoint [File Download]

Module 18 - Time Series Groups with Modeltime GluonTS

18.1 Data Collection & Initial Preparation

18.2 Deep Learning Models - DeepAR

18.3 Deep Learning Models - N-BEATS

18.4 Machine Learning - XGBoost

18.5 Evaluating Our ML & DL Models

18.6 Ensembles of ML & DL Models [The Best of Both Worlds] 🔥🔥🔥

18.7 DL Wrapup - Saving/Loading Models & Conclusions

CONGRATULATIONS!!! You. Did. It.

WOO HOO!!! Get YOUR Certificate & 25% OFF your next purchase! (1:07)