Working with real-time data streams in Python
summary
Claims that using a Kalman filter on-device yields far better live estimates than naive approaches, needing only two values of state and running in ~30–60 lines of Python from the core library. Demonstrations compare a window-5 and window-50 moving average to Kalman: data with a 20°→22° step over 100 samples shows the Kalman filter tracks the change faster with far less memory and avoids shipping raw streams (longer battery life, local control).
AI-written summary · about
chapters (11)
- 0:00Introduction to IOT
- 3:09The IoT Data Problem
- 5:06Centralized vs Local Data
- 7:52Real-time Data Challenges
- 9:44Introduction to Filtering
- 13:18Moving Average Filter
- 15:30Kalman Filter Explained
- 18:54Implementing Kalman Filters
- 22:38Filter Results Comparison
- 25:26Closing and Summary
- 27:01Q&A Session
related talks
47mIPython and Sympy to Develop a Kalman Filter for Multisensor Data Fusion
48mIdentification of Building Thermal Model - Using Data
32mEnabling real-time insights through stream processing in Python
34mMining smartphone sensor data with python
21mLeveraging Python for Real-Time Image processing
20mStreaming DataFrames: A New Way to Process Streaming Data
21mStreaming DataFrames: A New Way to Process Streaming Data
22mRunning, walking, sitting or biking? - Motion prediction with acceleration and rotation