mariofilho

7 ms·

1.
I Asked Gemini to Read 300-Year-Old Portuguese Parish Records(mariofilho.com)
9 points by mariofilho 2mo ago|0 comments
2.
A Kaggle Grandmaster Tries to Semi-Automate Himself(github.com)
2 points by mariofilho 7mo ago|0 comments
3.
by mariofilho 2y ago
I uploaded the image to Gemini 2.0 Flash Thinking 01 21 and asked: “ Here is a steps vs bmi plot. What do you notice?” Part of the answer: “Monkey Shape: The most striking feature of this plot is that the data points are arranged to form th
4.
A Game to Visually Understand Active Learning in Machine Learning(mariofilho.com)
1 points by mariofilho 2y ago|0 comments
5.
Kalman Filter for Time Series Forecasting in Python(forecastegy.com)
2 points by mariofilho 3y ago|0 comments
6.
How to Do Time Series Cross-Validation in Python(forecastegy.com)
2 points by mariofilho 3y ago|0 comments
7.
by mariofilho 3y ago
God forbid I put a link to my blog when sharing a useful tool for free!
8.
GPT-Summarizer: A Jupyter notebook to summarize long videos, podcasts, etc(github.com)
2 points by mariofilho 3y ago|2 comments
9.
Improving Training Stability for Multitask Ranking Models in Recommender Systems(github.com)
1 points by mariofilho 3y ago|0 comments
10.
TW-Bert: End-to-End Query Term Weighting(github.com)
1 points by mariofilho 3y ago|0 comments
11.
Unified Embedding: Battle-Tested Feature Representations for Web-Scale ML System(github.com)
1 points by mariofilho 3y ago|0 comments
12.
Why XGBoost Still Wins the Tabular Data Game(forecastegy.com)
3 points by mariofilho 3y ago|1 comments
13.
Does XGBoost Need Feature Scaling or Normalization?(forecastegy.com)
1 points by mariofilho 4y ago|0 comments
14.
Are Marketing Mix Models Useful? I Spent My Own Money to Find Out(forecastegy.com)
1 points by mariofilho 4y ago|0 comments
15.
How to Forecast Multiple Time Series at Once with Scikit-Learn(mariofilho.com)
1 points by mariofilho 7y ago|0 comments
16.
by mariofilho 8y ago
This is something I got curious about too. It's very likely the answer is no, but I would like to test it at some point.
17.
Can a Machine Learning Model Predict the SP500 by Looking at Candlesticks?(mariofilho.com)
28 points by mariofilho 8y ago|33 comments
18.
by mariofilho 8y ago
It will very likely fail to predict new data. The model can approximate the interactions over the range it saw on training, but we need to show it larger ranges if we want to predict for a bigger grid.
19.
by mariofilho 8y ago
Hi, thanks for your comment: I opened an issue on Github. This didn't seem obvious to many people that read the article, so it's nice that now we can keep this in mind while using the model: https://github.com/dmlc
20.
by mariofilho 8y ago
Hi, thanks for your comment. You are correct that this model will not generalize to numbers outside this range. My goal was just to have a reference when this questions comes up in a discussion about creating feature interactions that are a
21.
Can Gradient Boosting Learn Simple Arithmetic?(mariofilho.com)
47 points by mariofilho 8y ago|20 comments
22.
How to Use Neural Networks to Forecast Multiple Steps of a Time Series(mariofilho.com)
1 points by mariofilho 8y ago|0 comments
23.
How to Predict Multiple Time Series with Scikit-Learn (With an Example)(mariofilho.com)
3 points by mariofilho 9y ago|0 comments