Hurry! This Time Series Career Deal expires in:

How I Got 3 Raises In Two Years

And Kickstarted My Consulting Career

With Forecasting

Time Series Platinum

Time Series Platinum:

✅ High Performance Time Series ($897)
✅The Lost Time Series Modules ($499)
✅Auto-Forecast Shiny App + Lesson ($5,000)
✅2 Time Series Data Sets
($199)

Course Alone is $897

Unlock today for only $599 + Get 3 Bonuses

Sale Ends On August 12th

How I Got 3 Raises In Two Years And Kickstarted My Consulting Career With Forecasting


Dear Friend,

If you’re still using outdated methodologies like ARIMA, abandon them, and I’ll tell you why.

There’s a new way to do forecasts which can be done in minutes… instead of days.

This new method takes the routine of writing 10,000 lines of code, and chops it down to a measly 300…

… thereby cutting your work time to less than 1/10th of what it was… and accelerating the process 1000 fold.

Better yet, it’s at least 10% more accurate than any method available to the public as of now.

The bad news?

No one is teaching this technique online, or even in Universities.

When I graduated from college, I got a job at a company called Bonney Forge because I had a degree in mechanical engineering.

And while that’s a good field to get into, I wasn’t fulfilled.


What Bonney Forge did was manufacture valves and pipes for big manufacturers around the world. Which they obviously need Mechanical engineers for.

Now sure, the paycheck was steady, but really, I was just one of 1000 cogs in a big machine.

If layoffs were gonna happen… I’m in the crosshairs.

To make matters a little worse, I had just gotten married and had a daughter on the way.

Not wanting to be stuck at a measly $50,000 dollars a year for the rest of my life...

...I spent a lot of my time trying to develop new skills that I didn’t learn in college.

Eventually, I got good enough to get moved from the engineering team to the technical sales team for oil products.

Funnily enough, I had never done sales in my life, but what I was doing was acting as a business analyst… although I was still making the same pay as a mechanical engineer.

And In 2014 My Worst Fears Were Realized…


…The price of oil started dropping like a lead balloon.

What does that mean?

Well, at the time, the price of our products were largely based on the price of oil.

Which means when it was high, we could sell our products for a VERY high margin.

When it went down, we had to chop our prices down to stay competitive.

But if they went too low, we wouldn’t have any margin and would actually lose money on jobs.

Well, the price of oil went into an almost free-fall.

And it stayed that way for two years.

As the resident “Smart Guy” it was my job to solve this problem or get sacked.

So I had to buckle down and learn REAL Time Series.

First problem was most of the education available was outdated.

Learning data science is hard enough.

Being 10 years behind is even worse.

The second problem was the methods I was learning were too inefficient to do forecasts as frequently as I needed them.

Third problem was my company was losing money NOW and I needed to get these skills pronto.

That’s When I Decided To Reverse Engineer What Kaggle Winners Were Doing.


As you’re aware, the top data science teams compete in Kaggle, and one competition struck my eye in particular.

The 2014 Forecasting Competition for Walmart.

Walmart has tens of thousands of stores, multiple departments, 1000’s of sku’s and the winner was able to predict demand for all those products for all those stores.

So, I set my ego aside, and just studied what they were doing for weeks.

In the meantime, my division of the company was on fire.

Record layoffs.

Low workloads.

Even lower revenue.

I didn't know whether my division would be around in another quarter.

Not only did we have no idea how to fix the problem at hand, but quite frankly I was unsure of whether I was cut out for this job.

But after studying the best of the best... I finally built a working model to predict which jobs & products we could sell for a profit.

Because of that ONE model, I was able to take my division and steer them in the right direction.

Before Oil prices dropped, we were making $3,000,000 a year.

With my guidance, we were making $15,000,000 by 2016.

Not only did I avert the crisis, we were 600% MORE profitable.

And oil prices were still at a record low.

Despite Being The Manager For The Worst 2 Years Of The Company, I Had Gotten Promoted 3 Times.


By the end of 2016, I was reporting directly to the CEO, making $150,000 a year and went from a peon to the head honcho. (But that isn’t even the best part.)

Because of what I was able to do with forecasting, other companies asked me to train their data science teams too.

That added an extra 100,000 a year into my pocket and allowed me to train with people like S&P Global & Apple.


Not to mention that by the time my daughter turned 1, my wife was confident we were never going to run into money problems again.

Would You Like The Skills I Used To Do All This?


Listen, it's no secret as to why I was able to accomplish all this.

It's because I could predict what products we could sell for a profit, and put those jobs into production. That's entirely possible for anyone to learn how to do.

And back then? I was only scratching the surface of what was truly possible.

Now with advanced techniques like Feature Engineering, Machine Learning & Deep Learning, I could run circles around my former self.

And those are the advanced techniques I teach inside...

Time Series Platinum

CAREER DEAL


Time series platinum is the ultimate system that takes you from Joe Shmoe to Time Series Expert (and is YOUR secret weapon in the job market).

Inside Time Series Platinum, you gain my time series course + 3 bonuses to help you:

1. Become an Expert in Time Series (the guy or gal that YOUR company depends on to help them forecast)

2. Build an Interview Portfolio using Time Series (that will help you stand out from the crowd)

Let me explain...

Course #1:

High Performance Time Series

($897 Value)

Gain the forecasting skills to become the time series expert for your company.

HPTS Has 18 Modules & Is Divided Into 3 Sections:

Feature Engineering

Machine Learning

Deep Learning


"High Performance Time Series" is a masterclass in forecasting.

This class takes you from the very basic fundamentals (aka ARIMA) and teaches you even the most advanced techniques like Deep Learning.

Not only will you get to watch over my shoulder as I break down problems and give you the solutions...

... But I'm also going to walk you through each package step-by-step and show you both the right way and the wrong way to do things.

That way you'll not only know what to do, but also what NOT to do.

Let me take you through a quick walkthrough of the course.

Here's A Small Portion Of What's Inside:

  • Inject these 10 lines of code into your stack and create 20,000 forecasts in 30 seconds all without using a loop.


  • Want to stop taking forever to hyperparameter tune? Start your forecast with this overlooked feature to get a decent forecast from the jump. (This will increase your accuracy by 50% too.) 


  • The only algorithm I use when forecasting 100,000,000’s of data points (This algorithm cuts your production time by 90% and is still accurate at scale.)


  • Ever try to forecast with ARIMA and notice it’s incredibly slow? It’s because of this common mistake. Avoid it, and speed up your forecasts. 


  • What Rob Hyndman gets wrong about doing forecasts… and what he strangely gets right.


  • The backdoor way to hyperparameter tune large forecasts at the same time (thus cutting down your producing time by almost 75%!).


  • My 3 backwards techniques to train forecast models at scale. (The best part? These techniques virtually erase errors.)


  • Why Time Series Cross Validation shouldn’t be used all the time (plus what you should use instead.)


  • Discover the 6 waste of time algorithms that 99% of data scientists use. (And then they wonder why they have errors.)


  • Use this simple technique to maintain the order of splits when testing your models

  • My replacement for stationarity (In fact, I never use stationarity to forecast, and neither should you.) 


  • Discover the top 10 features that will increase the accuracy of your models for you

  • Steal my proven formula which determines your model’s accuracy to the Nth degree. (this formula will also show you where your forecast becomes inaccurate too.)


  • How to integrate Deep Learning to create automated forecasting models


  • A complete rundown of the modeltime forecasting workflow


  • This tuning technique replaces your error metrics & gives you a better analysis


  • Steal my “Dancho 12”. These are the Advanced Machine Learning Models I used to get a promotion to Director of Sales & Forecasting at Bonney Forge.

Here's Some Of What You'll Discover In

Part 1: Feature Engineering

💡The 5 Secrets of Time Series Forecasting Competitions Winning Solutions (See Module 0)

💡The 6 most important time series visualizations and diagnostics (See Module 2)

💡Summarization, padding time series, filtering time, joining external regressors, and extending time series into the future (see module 3)

💡Time series transformations including log, box cox, rolling windows, smoothing with Loess, normalization and standardization, imputation and outlier cleaning, lagging and differencing, and fourier transformation (see module 4)

💡Find outliers and apply treatment techniques (like imputation) (see module 4)

💡Feature engineering: Calendar and time-based features, splines, time-based interactions, Fourier features, lag features, special events, and using Recipes & Timetk (Modules 5 & 6)

Here's Some Of What You'll Discover In

Part 2: Machine Learning

💡Learn the modeltime workflow so you can use the 5 functions to produce a full forecast, and also learn additional features that modeltime offers to make your forecast organized & accurate (page 2 of cheat sheet, Module 7)

💡ARIMA, Prophet, Exponential Smoothing, TBATS, Seasonal Decomposition (See Modules 8, 9, & 10) - Know your enemy (slow speed, tend to be less accurate, need 10,000 models to predict 10,000 time series)

💡Machine Learning - XGBoost, Random Forest, Elastic Net, Support Vector Machine, Neural Net (High accuracy & Fast, 1 model predicts all 10,000 time series) (See Module 11) (<- 4 hours long)

💡Boosted ARIMA and Boosted Prophet - Improve ARIMA and prophet models by using XGBoost to mine the modeling error for additional pattern

💡Hyperparameter tuning and cross validation to stabilize machine learning model performance and confidently forecast the future (Cheat Sheet Page 3, Module 13)

💡Ensembling multiple models - when you combine the strength tend to help smooth out and stabilize the ensemble model. Average, Weighted Average, Stacking & Multi-Level Stacking (Module 14)

💡Forecasting at scale - Forecast 20 time series at once with 20 different models. Find the best, cross validate, ensemble, and forecast the future for 20 time series at once with one final ensemble model. (all in seconds)

Here's Some Of What You'll Discover In

Part 3: Deep Learning

💡Use Modeltime GluonTS to create a single DeepAR model that generates 20 forecasts (Module 17)

💡Use both Machine Learning & Deep Learning to forecast 20 time series (Module 18)

Here’s What Other Students Have Said

After Taking The Course:

Jaimie created used the time series course to improve revenue forecasts.

Graeme got a 20% pay raise by applying time series course to his company's forecasts.

Amit got a new job as Machine Learning Associate with the time series course.

Matt got 2 competing job offers and secured a new job as a Lead Data Scientist using the time series course.

Bonus #1:

The Lost Time Series Modules

($499 Value)

The Official Expansion Pack To High Performance Time Series

The Lost Time Series Modules focus on the 5 advanced skills I've mastered since making the Time Series Course:

▶︎ Forecasting in Changing Demand (supply disruptions, COVID, and recessions)

▶︎ Iterative Forecasting

▶︎ Recursive Iterative Forecasting

▶︎ Automatic Machine Learning with H2O

▶︎ Hierarchical Forecasting

Recipe 1: Forecasting In Changing Demand

  • Unlock 4 strategies for forecasting recession, COVID impacts, and supply disruptions.

  • Discover the secret to using Deep Learning (and why it beats most Machine Learning models in supply disruptions)

  • Uncover the hack to making a dozen XGBoost models in 4 lines of code.

  • Gain a new modeltime feature that increases confidence interval accuracy by up to 43%.

  • Use 15 simple (but effective) models to forecast short time series (in 5 lines of code).

  • Discover the secret to stabilizing forecast models in the presence of changing demand (hint: we ensemble these 3 types of models)

  • Produce optimal forecasts at scale (and know the true probability of being correct).

  • Compare 17 models (and discover the trick to picking the best for each time series in your data).

Recipe 2: Iterative Forecasting

  • Uncover the sneaky-easy technique for making 100's of iterative forecasts (without writing a single For-Loop).

  • Find out the dead-simple way to clean and prep your time series data for iterative forecasting (in only 3 lines of code).

  • Unearth the NEW easy way to hyperparameter tune your models inside of the iterative forecasting workflow (I've never taught this before).

  • Use this special technique for handling errors inside of the iterative forecasting workflow.

  • Gain 2 brand new modeltime models that almost no one knows exist (and they beat XGBoost 4 out of 5 times on this time series dataset).

  • Discover the trick to picking the best forecast models for each time series in only 1 line of code.

Recipe 3: Recursive Iterative Forecasting

  • Uncover the secret to including short-term lags into your forecasting models (an essential skill for forecasting monthly time series)

  • Discover the most important concept to making recursive models inside of modeltime (this skill alone will make your autoregressive models 1000X better)

  • Identify 2 key features you can include inside of your forecasts

  • Unlock Transformer Functions (that increase autoregressive model performance 90% of the time).

  • Find out how to gain additional performance in 10 of 18 models (56% better models) with this one approach.

  • Discover how to add External Regressors, Lags and Engineered Features inside of the Iterative

  • Forecasting Workflow (I've never taught this before).

  • Visualize 18 forecasts at once with this NEW feature in Plot Time Series (without filling your entire screen with plots).

Recipe 4: Automatic Machine Learning

  • Get my simple method for preparing time series for Automatic Machine Learning in 3 lines of code.

  • Make 100+ models in under 5 minutes automatically (This work is fully automated so you can grab a coffee AND make models at the same time).

  • Complete forecasts that used to take me over 1 week in under 15-minutes (and in 15% of the code).

  • Uncover the special trick for handling H2O models inside of the Modeltime Forecast Workflow (this makes saving and loading a breeze - a MUST for production).

  • Find out 3 uses for H2O's New Variable Importance Feature (one of these can actually improve your model even more!).

  • Discover a secret between Mac's and PC's that will make you a Mac/Linux lover (and improve your forecasts).

Recipe 5: Hierarchical Forecasting

  • Unearth the dead-simple way to adapt Hierarchical Time Series data for forecasting (You won't believe how much time these 10 lines of code saves you).

  • Gain a deep-dive into the 2 most popular algorithms for Hierarchical Time Series Forecasting at Scale.

  • Find out the 1 parameter that can instantly increase the performance of your boosted hierarchical time series models.

  • Expose how baseline models can be used to not just compare performance BUT actually improve performance.

  • Use a simple (but effective) strategy to combine multiple models into one for Hierarchical Time Series Forecasting.

  • Shrink Hierarchical Forecasting error using this key technique I learned from the M5 Competition (It improved my forecast accuracy 37%)

  • Use a new window algorithm that as many as 98% of data scientists aren't using (it's fast and useful for Hierarchical data).

  • Use this secret post-processing technique to incorporate Business Knowledge into your forecasts (and immediately improve your results 8%).

Want Help With Your JOB PORTFOLIO?

Then PAD your PORTFOLIO with these.

(Read On)

👇


Bonus #2:

Auto-Forecasting Shiny App + Training

(Shiny App That's Worth $5,000)

Would You Like Legitimate Business Tools That Companies Will Hire YOU For?

What do you think is going to happen when you add this app to your portfolio?

Well... Here's what happened to Jennat.

Jennat was interviewing for data science positions.

She walked into the Bank Of Canada (the backbone of Canada's financial system) for her interview.

Everything was against her.

The team used Python.

And she was an unknown.

The manager took one look at her project portfolio.

Jennat says, "The SUPRISE is that when they saw a Shiny App forecasting using Modeltime, they created a higher position for me!"

Jennat was literally promoted in the interview.

And what set her apart was the Shiny App that Forecasted with Modeltime!

Unleash the power of modeltime + timetk with an automatic forecasting app that can handle a wide range of time series, and accurately predict the future.

You Get MY EXACT Shiny App I made that Jennat used to land her dream job!

(Add it to your portfolio & watch as your interviewer gleans with excitement when you show it off)

And Gain A Full 60-Minute Training On The Apps Secret Sauce

  • Steal my Nostradamus Lite Shiny App (I've sold lesser apps for $5,000+)

  • Automatically produce a forecast for as many time series as you need (with only 1 model).

  • Get a full 60 minute training on how the algorithms (including modeltime and timetk) work inside of the shiny app

  • Avoid a key mistake when preparing never-before-seen time series (that cost me hours of rework)

  • Uncover the hack to combining multiple models into one super learner.

  • Use a potent blend of models to increase accuracy 15%.

  • Combine 5 machine learning models in 3 lines of code.

  • Discover my secret to engineered features for automatic forecasting

  • Learn the trick to making forecasts on-demand inside of a Shiny App (without taking a long time).

Bonus #3:

2 Time Series Data Sets

($199 Value)

Build Your Portfolio While You Practice Your Skills

What's the one thing that my students ask for when they purchase my courses?

Answer: MORE DATA.

Why? It's because they want to practice their skills as they learn in my courses AND while they make their portfolios stand out.

Want to kill 2 birds with 1 stone?

I'm giving you 2 unique time series data sets for you to practice your skills + make a killer, career accelerating interview portfolio so you can stand out from the crowd:

Data Set #1: Stock Prices

The First Data Set is Daily Stock Prices for 49 of the largest companies in the world.

✅ 39,347 Rows by 8 Columns

✅ 49 Time Series Groups

✅ Daily Interval

✅ Missing Time Stamps (Weekends & Holidays)

Data Set #2: US Energy Production

The 2nd Data Set is Monthly Energy Production for 49 of the largest companies in the world.

✅ 7,631 Rows by 5 Columns

✅ 20 Time Series Groups

✅ Monthly Interval

✅ Production history going back to 1973

Time Series Platinum

  • High Performance Time Series ($897)

  • The Lost Time Series Modules ($499)

  • Auto-Forecasting Shiny App + Lesson ($4,997)

  • 2 Time Series Data Sets ($199)

Total Value = $6,592

The Course Alone is $897

Yours Today For $599 + 3 Bonuses

Frequently Asked Questions

Q1: How long do I have access?

You get lifetime access to the courses, 3 bonuses, and BSU Community Slack Channel for support.

Q2: I'm a beginner. What skills do I need to start?

This course will teach you time series. We do have another course, Data Science for Business Part 1 that teaches foundational data science skills (with no prior experience required).

Q3: How long does it take to complete the courses?

30 to 60 Days for the High-Performance Time Series Course with on average 10 hours per week.

Q4: Am I guaranteed a job?

I cannot guarantee anything. I will do everything in my power to help you get a job. But effort and motivation are up to you (out of my control). And those are the leading indicators of success.

Q5: Will this course teach me data science skills too?

This course is focused on Time Series. We do not teach data science fundamentals or web apps. I have 4 other courses in the 5-Course R-Track that dive deep into those topics.

Q6: Do you offer a money-back guarantee?

Yes. We offer a 30 day moneyback guarantee. If you are unhappy please contact us in 30 days for a full refund. See our refund policy for more details

Q7: Is this course for R-users?

This course is designed for R-users specifically. While the broad concepts can be used for Python-users, the code in the course is written with R and Rmarkdown. So it may feel weird if you don't know R.

Q8: What about support?

By enrolling in this course, you gain lifetime access to our Private Slack Community. Mentorship and guidance is provided virtually through the Business Science Slack Community.

Time Series Platinum

  • High Performance Time Series ($897)

  • The Lost Time Series Modules ($499)

  • Auto-Forecasting Shiny App + Lesson ($4,997)

  • 2 Time Series Data Sets ($199)

Total Value = $6,592

The Course Alone is $897

Yours Today For $599 + 3 Bonuses