Master Time Series Forecasting in 60 days
Develop a career-boosting skill without putting your life on hold
Let's go!
Are you tired of:
😤
Reading incomplete time series tutorials on Medium where you can't even reproduce the results?
🤬
Browsing time series courses on Coursera only to find out that they are full of theory and exercises are in R?
😫
Browsing graduate courses that would force you to work part-time and spend thousands of dollars?
Don't worry, I was in the same place
As a professional data scientist, I use Python daily, but for some reason, everything about time series is in R!
I tried many courses, but all were too theoretical with manual derivations of equations. I don't want to prove a theorem, I want to solve real-life business problems!
Plus, I didn't have time or the money to go back to school and take a graduate program to learn time series forecasting.
Now, imagine this:
✅ Having a step-by-step process to master time series forecasting using 100% Python
✅ Being able to master time series forecasting in 60 days, on your own time, without putting your life and work on hold
✅ Completing real-life projects to add you data science portfolio
⭐⭐⭐⭐⭐
"As a professional data scientist, I was looking for a way to learn time series because I need it in my job. I had very limited knowledge on this subject. With this course, I learned everything I needed and it helped me a lot to succeed in my project. Marco is a great teacher! Thank you! I recommend it to everyone who wants to learn about time series!"
Askia Khalid
Data Scientist
Introducing 🥁
Applied Time Series Forecasting in Python
The only online course that combines both statistical and deep learning methods for time series forecasting, all in Python.
⏩
Fast path to mastery
Get on the fast track to master time series forecasting, from the basics to the latest techniques!
🐍
100% Python
The course uses only Python and TensorFlow! Use Jupyter or your favorite IDE as you follow along!
🔨
Project-based
I believe in learning by doing! Complete more than 16 end-to-end forecasting projects throughout the course!
Meet your instructor 👋
Hey, my name is Marco and I am your instructor and founder of Data Science with Marco.
My Background
I studied chemical engineering, but let's just say that wasn't my passion! I started learning web development on my own, landed my first job in the field, and studied data science at night. Self-learning data science was tough. I focused a lot on practical skills and building a portfolio of projects, which led me to my current role as a senior data scientist at one of Canada's largest banks.
Learning Time Series
There is a lot of time-dependent data in a bank, so I felt the need to learn time series forecasting. I had to translate R code to Python, debug bad online tutorials, and piece together many blog articles. The experience was so painful that I started sharing my learning through blog posts to make it easier for others to learn this new skill.
This led me to publishing a book with Manning Publications on time series! But I still had more to say on the subject, so I made this course! The absolute best way to learn time series forecasting. A complete course, easy to follow, with end-to-end projects, and 100% Python code.
Inside the course 🔎
Module 1: Introduction
We kick off the course with a gentle introduction to time series. We define them, explore their components and understand why there is a unique way to work with time series data. Then, we start forecasting with baseline models
Project 1: Applying baseline models to forecast milk production
Module 2: The random walk
The random walk is a special case where we cannot make reasonable forecasts! Nonetheless, it is super important to understand the random walk process, as we explore fundamental concepts such as stationarity and autocorrelation.
Project 2: Analyzing the daily closing price of Amazon (AMZN)
Module 3: Forecasting with the SARIMAX family of models
Things get interesting as we start working with the SARIMAX family of models! We first master the fundamental models AR(p) and MA(q), before combining them into the ARMA and ARIMA models. Then we add the ability to model seasonality with SARIMA, and add exogenous variables to our model using SARIMAX. We also design a general modeling procedure to be applied in any situation.
Project 3: Forecasting a simulated MA process
Project 4: Forecasting a simulated AR process
Project 5: Forecasting a simulated ARMA process
Project 6: Forecasting quarterly electricity production with ARIMA
Project 7: Forecasting monthly milk production with SARIMA
Project 8: Forecasting monthly price of cattle with SARIMAX
Module 4: Multivariate forecasting
In this module, we see how we can forecast more than one time series at a time, with the models VAR, VARMA and VARMAX. We also explore the concept of Granger causality, which is the basis of these models.
Project 9: Forecasting monthly price of livestock in Saskatchewan
Module 5: Exponential smoothing
Exponential smoothing is a forecasting procedure that is very fast, flexible and powerful. In this module, we slowly build from simple exponential smoothing, to double and triple exponential smoothing.
Project 10: Forecasting monthly corticosteroid drug subsidy
Module 6: Dealing with multiple seasonal periods
Working with a single seasonal period is easy. But what if there is more than one? What if your data has a daily and a weekly seasonality? Then, we need to apply models like BATS or TBATS.
Project 11: Forecasting hourly traffic on the interstate
Module 7: Forecasting using decomposition
A unique forecasting approach where we forecast long-term and short-term effects and then combine them for a final prediction. This is done through the use of the Theta model.
Project 12: Forecasting weekly CO2 concentration
Module 8: Deep learning for time series forecasting
In this module, we apply different deep learning models for forecasting time series data. We implement a robust framework for any situation: single-step, multi-step or multivariate forecasting. We also implement early stopping and learning rate scheduling for the training of our models.
Project 13: Forecasting hourly load of an electricity transformer (ETT dataset)
BONUS 1: Prophet
Prophet is a popular forecasting library open sourced by Meta (formerly Facebook). This forecasting procedure is powerful and very easy to use. Knowing how to work with Prophet makes it very easy to work with any automated forecasting library.
Project 14: Forecasting hourly electricity consumption in a household
BONUS 2: State-of-the-art time series forecasting
In this module, we explore and implement the latest advances in time series forecasting. This module will be updated as new methods are designed and made available. For example, we explore N-BEATS, N-HiTS, PatchTST, TimesNet, TimeGPT and more.
Project 15: Forecasting daily minimum temperature
BONUS 3: Intermittent time series forecasting
Your series can sometimes have long periods of zero values, making them intermittent. For those, we need specific models, like Croston's method, ADIDA, IMAPA and the TSB model.
Project 16: Forecasting the demand of car parts
What my students say 💬
⭐⭐⭐⭐⭐
"This course is really what I searched for: an intermediate course for time series. It's a good way to start with the review of the basics, because afterwards, everyone is on the recommended level. I would do another course from this instructor immediately. Nice job!"
Luis Kalckstein
Data Scientist
⭐⭐⭐⭐⭐
"This is an amazing video series. The instructions are clear and concise, and the instructor is very knowledgeable. Great course."
Oscar Paulse
Data Scientist
⭐⭐⭐⭐⭐
"This course goes straight to what you need to do to get things done! Pretty hard to find it out there."
Filipe Schenkel de Souza
Data Scientist
⭐⭐⭐⭐⭐
"Very thorough and touches everything of importance."
Shankar Viswanath
AI/ML Developer
⭐⭐⭐⭐⭐
"Clear explanations, great content, perfect pace."
Josip Nemet
Senior Business Manager
⭐⭐⭐⭐⭐
"The instructor is an expert in time series concepts in Python. Must follow this course."
Mayurkumar Surani
Data Scientist
Enroll risk-free today!
I am convinced that you will love this course. Not entirely satisfied? You have 30 days to get your money back!
