On My Mind: MicroPython, Python, and Data Science

Imagine if Python went on a diet, ditched its fancy clothes(libraries), and decided to live in the minimalist world(of tiny microcontrollers). The lean, mean, code running machine that fits in your pocket (literally). While everyone else is out here training colossal AI models and holding terabytes of data, MicroPython, on IoT devices, in smart homes, and on drones; quietly collecting the data and powering the big leagues.

In this blog, I will take you on an eye-opening journey into the world of MicroPython, how it’s sneaking its way into data science, how it differs from Python, and why you should care about this pint-sized coding hero.

So, if you have ever wondered how your smartwatch knows your heartbeat, or how that robot vacuum knows where not to vacuum your sleeping dog, chances are MicroPython is somewhere. It’s Python, but for devices that don’t have room for all the bells and whistles.

I know what Python is but MicroPython?

MicroPython is a lightweight implementation of Python 3 which is designed for microcontrollers and embedded systems. It is optimized to run on resource constrained devices, which have limited RAM and CPU power. MicroPython is compact, efficient, and comes with modules that provide access to hardware level functionalities like GPIO (General Purpose Input/Output), I2C, SPI, and UART and hence making it ideal for the Internet of Things (IoT) and robotics applications. It is not "so-old" software but was developed in 2013 and written by Damien George in C language.

MicroPython’s simplicity and resemblance to Python have made it a favorite among professionals in the IoT domain. It allows developers to write Python code that directly interfaces with hardware components like sensors, actuators, and communication modules. MicroPython simplifies the development work for hardware projects that were previously dominated by C or C++.

Similarities between Micropython and Python:-

1. Both MicroPython and Python are dynamically typed, meaning variables don’t need explicit type declarations. Example: x = "This line is dynamically defined here and it is now a string!"
2. Common Python data types like: Numbers (int, float), Strings, Lists, Dictionaries, and Tuples are all supported in MicroPython.
3. Both support defining functions and creating classes for object-oriented programming.
4. Both provide an interactive REPL(Read-Eval-Print Loop) environment for testing and executing commands line by line.
5. Standard Python Features, like Conditional Statements (if, elif, else), Loops (for, while), Error Handling (try, except, finally) and function definitions are present in MicroPyton too.
6. MicroPython uses a subset of Python's module system that allows users to import functionality. Common modules like math, random, and time are available in both environments.
7. Both MicroPython and Python use automatic memory management via garbage collection to clean up unused memory.

How MicroPython Is Not "Just Python":-

The primary goal in the making of MicroPython was to adapt Python's environments for hardware where constraints are faced in terms of memory, processing power, and storage is tight. While MicroPython’s origins lie in Python, it diverges in many ways:-

• Some Python features, like full os or math modules, are trimmed/simplified.
• MicroPython includes unique modules like machine and pyb which support hardware functionalities(for controlling GPIO pins, I2C, SPI, and others) that don’t exist in Python.
• MicroPython runs on microcontrollers without an operating system, whereas Python typically relies on a desktop OS.
• It has also removed unsuitable(large libraries) features for microcontrollers due to their size, like NumPy.
• MicroPython has replaced complex modules with lightweight(hardware-specific ones) (e.g., machine and network).
• MicroPython has a custom Python interpreter that runs efficiently on devices with as little as 256KB of memory.
• Unique modules, such as 'machine' for hardware interaction, were created to replace Python's 'os' and 'sys' modules, which are unsuitable for embedded systems.

Exploring the Intersection of MicroPython and Data Science:-

MicroPython is indirectly supporting data science. Data collection and preprocessing are undetachable and critical steps in the data science, are now being offloaded to resource constrained devices at the edge.

The intersection of MicroPython and data science lies in its ability to:

1. IoT and Data Collection

MicroPython is deployed on microcontrollers (e.g., ESP32, ESP8266) to collect real world data from sensors (e.g., temperature, humidity, motion). This data becomes the raw input for downstream data science analysis.

Example: A smart agriculture project uses MicroPython to gather soil moisture data, which is then processed using Python based machine learning algorithms to predict irrigation needs.

2. Edge Computing and Preprocessing

In edge computing setups, data preprocessing (e.g., filtering, normalization) happens on MicroPython powered devices before sending data to the cloud for advanced analysis.

Example: A wearable device running MicroPython calculates real time heart rate averages and only transmits anomalous patterns to a central data processing system.

3. Prototyping Data Pipelines for Embedded Systems

Data scientists and engineers use MicroPython to prototype IoT devices that collect data for real world machine learning applications.

Example: An environmental monitoring station uses MicroPython to collect air quality metrics, which are analyzed later to predict pollution trends using a Python based ML model.

4. Custom Data Devices

MicroPython is used in projects requiring specialized hardware setups that are tailored for specific data science problems.

Example: Building a low cost spectrometer powered by MicroPython to collect spectral data for chemical analysis.

Challenges of Using MicroPython in Data Science:-

1. Microcontrollers have limited memory and processing power.
2. No direct support for data science libraries like Pandas or NumPy.
3. It is best suited for data collection and preprocessing rather than full-fledged analysis.

The original full source code is available on Github:-

https://github.com/micropython/micropython