You can use tsfresh relevance table to solve this issue. After you extract your features with tsfresh:
from tsfresh.examples import load_robot_execution_failures
from tsfresh import extract_features, select_features
from tsfresh.feature_selection.relevance import calculate_relevance_table
y = pd.Series(data = extracted_features['class'], index=...
There are two end of the spectrum:
Write a SQL query that creates a single materialized views that formatted in such a way that it is ready for machine learning.
Write several SQL queries that fetch all possibly relevant data from the database. Then munge them in another system to create a table ready for machine learning.
It is often times the ...
You are describing different coordinates but suppose for a second that points are represented as cartesian coordinates $(x,y,z)$. A surface consists of infinitely many points which cannot be stored by a computer. One solution to this problem is that we put a grid over the $(x,y)$-plane and store for each point in the grid the height value $z$. Here, is an ...
A brute force approach has O(n) search complexity no matter if you do it in Python or a database. For faster queries you need a tree-structured multidimensional lookup table, for example, a k-d tree. For Python, there are implementations of a k-d tree in both SciPy and Scikit-Learn:
One option is a two step process:
Fit spectral clustering to find the groups. Since spectral clustering is graph based, it is good at finding connected groups.
For each cluster, individually estimated the bounded regions of the planes. This is often called convex hull.