Meteorite-Mania
feature-engineering-tutorials
Meteorite-Mania | feature-engineering-tutorials | |
---|---|---|
1 | 1 | |
1 | 266 | |
- | 1.1% | |
10.0 | 0.0 | |
over 1 year ago | 2 days ago | |
Jupyter Notebook | Jupyter Notebook | |
- | GNU Affero General Public License v3.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
Meteorite-Mania
-
Meteorite Mania
If you want to find out more about how I put this together (including seeing some fun exploratory visualizations) you can view the GitHub Repository for this project. Please feel free to drop a comment or reach out via LinkedIn.
feature-engineering-tutorials
-
How to balance multiple time series data?
I’ve actually solved a similar problem several times in a variety of settings. I’ve had success with boosted trees and feature engineering on the sensor readings over time. I treat each reading as an observation and set the target to be the value I want to forecast (e.g. one hour ahead, the sum over the next day, the value at the same time the next day). There was a recent paper that compared boosted trees to deep learning techniques and found the boosted trees performed really well. Next, I perform feature engineering to aggregate the data up to the current time. These features will include the current value, lagged values over multiple observations for that sensor, more complicated features from moving statistics over different time scales, etc. I actually wrote a blog about creating these features using the open-source package RasgoQL and have similar types of features shared in the open-source repository here. I have also had success creating these sorts of historical features using the tsfresh package. Finally, when evaluating the forecast, use a time based split so earlier data is used to train the model and later data to evaluate the model.
What are some alternatives?
jupyter-notebook-chatcompletion - Jupyter Notebook ChatCompletion is VSCode extension that brings the power of OpenAI's ChatCompletion API to your Jupyter Notebooks!
intro-to-python - An intro to Python & programming for wanna-be data scientists
dtreeviz - A python library for decision tree visualization and model interpretation.
ydata-quality - Data Quality assessment with one line of code
gastrodon - Visualize RDF data in Jupyter with Pandas
PRML - PRML algorithms implemented in Python
desbordante-core - Desbordante is a high-performance data profiler that is capable of discovering many different patterns in data using various algorithms. It also allows to run data cleaning scenarios using these algorithms. Desbordante has a console version and an easy-to-use web application.
FeatureHub - The most comprehensive library of AI/ML features across multiple domains. Our goal is to create a dataset that serves as a valuable resource for researchers and data scientists worldwide
jupytemplate - Templates for jupyter notebooks
PyImpetus - PyImpetus is a Markov Blanket based feature subset selection algorithm that considers features both separately and together as a group in order to provide not just the best set of features but also the best combination of features
pandas-profiling - Create HTML profiling reports from pandas DataFrame objects [Moved to: https://github.com/ydataai/pandas-profiling]
ydata-profiling - 1 Line of code data quality profiling & exploratory data analysis for Pandas and Spark DataFrames.