The micro:bit machine learning tool lets you
train, test and improve a machine learning
model that can recognise different actions,
or movements.
Here’s a quick tour of how it works.
You’ll need two micro:bits.
You’ll hold or wear the first micro:bit.
Follow the instructions on screen for setting
up micro:bit 1.
Then unplug it and attach a battery pack.
If you have a wrist strap, you can attach
micro:bit 1 to your arm.
The second micro:bit stays connected to your
computer with a USB cable.
Follow the instructions on screen for setting
up micro:bit 2.
At the bottom of the screen you’ll see live
movement data from micro:bit 1’s
accelerometer sensor.
Each coloured line represents a different
direction, or dimension, you’re moving the micro:bit in.
Step 1, add some data!
Decide which actions you want the micro:bit
machine learning tool to recognise.
Clapping and waving are good ones to start
with.
Name your first action, then start moving!
Click the red button to collect your first
sample of data.
Collect at least three samples
of your first action.
And do the same for at least one other action.
Step 2, train the model.
This means that the machine learning tool
makes a set of rules to estimate which action
you’re doing when moving micro:bit 1.
Step 3, test the model.
Try making each of your actions.
The screen shows which action the model estimates
that you’re making.
The percentage number is how confident, or
sure, the model is.
Now try it yourself and see what different movements you can train the micro:bit machine learning tool to recognise.