In Bakhmut a restaurant owner and 60,000 convicts destroyed an army that NATO trained for 9 years. Reminds me of the farmers in sandals that kicked the US out of Afghanistan. The US Govt spends a trillion dollars >per year< on a military that can’t win anything. What a waste.
When it comes to forecasting:
- R is better than Python
- Deep learning rules supreme
- Facebook Prophet is the number 1 forecasting model
- Bayesian methods always produce perfect forecast even if you can’t specify correct prior. And if you really need a prior just ask ChatGPT to produce one
- One doesn’t need much data for forecasting, a few points are enough if you pick the best transformer
- No one needs probabilistic forecasting, point forecast is all one needs
- Forecasting stock prices using Prophet is the way to go
- Learning about #timeseries and #forecasting is boring; just ask ChatGPT to produce your forecast
- Forecasting has no business value whatsoever
- The best way to land data science job is to spin LSTM on 30 data points
- Transformers rule supreme. They always win, just ask hundreds of transformers paper
- SAP is a forecasting powerhouse; nothing ever comes close
- time series is like language, because one can replace the worlds with numbers. Hence transformers are the way to go
- Exponential smoothing was invented in 2023
- Deep learning won the M5 forecasting competition and also won every single forecasting completion before that, starting with M1 in 1982
- One can predict stock and bitcoin prices using ARIMA, even if it is ARIMA(0,1,0)
- MAPE is the best forecasting metric ever
- benchmarks don’t matter, not at all and if transformer can’t beat your benchmark you should have not used benchmark in the first place
- Zestimate is the most accurate model in the history of house price prediction
- Forecasting Value Added is non starter, don’t use it it is a waste of time and space
- the best forecasting data scientists live in Colorado and they use Facebook prophet. When they don’t forecast house prices for Zillow they moonlight writing medium articles how to predict stock prices and make you rich
#deeplearning #learning #timeseries #forecasting
ChatGPT is overhyped.
That's what I told myself after 2 weeks of trying (and failing) to use it well.
Turns out, I was just a poor prompt writer.
But after spending hundreds of hours tinkering, I've finally cracked it.
And now, it's my personal writing assistant.
Here's how:
[NEW] Machine Learning Specialization - Andrew Ng
ML specialization is the revamped edition of the former Stanford machine learning, one of the famous and earliest AI courses. Materials for the new course are now freely available. No Matlab, just Python!
https://t.co/ST5VWkoboj
11 ways ChatGPT saves me hours of work every day, and why you'll never outcompete those who use AI effectively.
A list for those who write code:
1 of 16
I'm reading a book a week in 2023. Classics, sci-fi, nonfiction, or anything people highly recommend. I'll keep adjusting the list. Start on Monday, done by Sunday. Might make lowkey videos of takeaways. If you want to read along, the current list is here: https://t.co/hYCs1pvLiZ
The sharpest minds on Wall Street predicted the S&P500 for 2022
Goldman Sachs - 5,100
Wells Fargo - 5,200
JP Morgan - 5,050
Finished at 3,839
Not one close
Pay zero attention to forecasters. ZERO