Name what you already understand before the build gets bigger.
Applied data
Basic Statistics for Sensor Logs
Turn repeated maker readings into a small evidence note: typical value, noise width, suspicious jumps, and what to measure again.
Topic goal to ladder route
Know the destination, then climb the route.
A topic is the maker goal. A ladder is the route from what you understand now to one visible proof you can build, sketch, test, or explain. This one ties back to Build an Obsidian project notebook.
Read a short sensor log and explain mean, median, range, standard deviation, outliers, and moving-average tradeoffs in project terms.
Read the short lesson, watch one useful source, sketch the idea, check the math, then practice.
Take ten sample readings from a sensor, RSSI sweep, temperature probe, or simulated list. Write the mean, median, min, max, range, one suspicious reading, and the next repeat test.
Source tutorials for sensor data
Use the video as source material for notes, cards, and practice. The written ladder still works without playback.
Use the controls to compare source tutorials. The first card embeds a privacy-enhanced player; alternate cards open on YouTube so the page stays fast.
1 of 3
Primary source
Calculating the Mean, Variance and Standard Deviation, Clearly Explained
Video by StatQuest with Josh Starmer · Open on YouTube
A clear statistics source for turning a messy sensor log into mean, spread, and outlier questions before making claims.
First watch: Watch for the difference between average value, variation, and why one reading is not enough.
- Mean
- Variance
- Standard deviation
- Interpreting spread
Practice after watching: Take ten readings from a sensor log and write the mean, min, max, and spread before deciding whether the signal changed.
Preference saves on this device.
Open on YouTube
Alternate source
Arduino Moving Average Filter for ADC Noise Reduction
Video by Abdul Rehman · Open on YouTube
Useful maker-scale source for seeing why repeated analog readings need smoothing and why the raw stream should still be inspected.
First watch: Watch for the raw readings, averaging window, and what the smoother hides or reveals.
- Raw ADC noise
- Averaging window
- Smoothed output
- Tradeoff
Practice after watching: Compare raw values and a 5-sample moving average, then write what changed and what lag was introduced.
Preference saves on this device.
Open on YouTube
Alternate source
Kalman Filter for Beginners Explained: Recursive Filters and MATLAB
Video by MATLAB · Open on YouTube
A more advanced source for learners who already understand mean and spread and want to see why filters estimate state from noisy measurements.
First watch: Watch only for the measurement-vs-estimate idea first; do not jump to a Kalman filter before the simple log checks make sense.
- Noisy measurement
- Recursive average
- Moving average
- Estimate vs raw signal
Practice after watching: Write when a simple moving average is enough and when a project would need a more explicit state estimate.
Preference saves on this device.
Sensor log check
Do not trust one reading. Inspect the shape of the readings.
A maker sensor log should answer three questions before a decision: what is typical, how wide is the noise, and which readings deserve a second look.
Typical value
Mean or median: the value your project usually sees when the setup is stable.
Noise width
Range or standard deviation: how much the reading moves even when nothing important changed.
Suspicious reading
Outlier: a jump that might be a real event, bad wiring, timing, or a measurement mistake.
Smoother tradeoff
Moving average: reduces jitter, but it can lag behind fast changes.
Ladder steps
Each step should prove one idea before the project asks for the next one.
Name the measurementA number is not useful until the unit, sensor, sample rate, and setup are named. Your note says what was measured and what changed during the sample.
Find the typical valueMean or median gives a center to compare against new readings. You can say whether the value is typical or unusual for this setup.
Measure the noise widthRange and standard deviation show how much the readings move when the project may be stable. Your note names the normal wiggle before calling something a signal.
Mark suspicious jumpsOutliers can be real events, wiring problems, timing issues, or sensor glitches. Your note marks the jump and names the next repeat check.
Smooth only after inspectionA moving average can reduce jitter, but it can hide fast changes or add lag. Your note says what the smoother improved and what it might have hidden.
Examples to inspect
Use examples to read signals, not as blind recipes.
Spot a suspicious jump
Project signal
readings = [22.1, 22.4, 22.2, 28.9, 22.3]
Expected signal: 28.9 is not automatically wrong, but it deserves a repeat check
Caution: Do not delete an outlier before explaining why it does not belong.
Summarize the log before judging it
Project signal
mean, median, min, max, range
Expected signal: The summary separates center and spread from one dramatic value
Caution: Keep units attached to every number.
Smooth jitter after reading raw values
Project signal
moving_average(window=5)
Expected signal: The line gets calmer but reacts more slowly
Caution: Do not smooth away the event you are trying to detect.
Self-check: can you use this?
Answer these before the practice task. The quiz checks your answers on this page only; nothing is saved.
0 of 8 checked.
Common traps
- Deleting outliers because they are inconvenient.
- Averaging readings that use different units or setups.
- Publishing a smoothed graph without keeping the raw values.
- Calling noise a trend after one short sample.
- Using a moving average without naming the window size or lag.
Practice task
Take ten sample readings from a sensor, RSSI sweep, temperature probe, or simulated list. Write the mean, median, min, max, range, one suspicious reading, and the next repeat test.
Next steps
- Use this before deciding whether an RSSI direction peak is stable.
- Use it before filtering ESP32 or Arduino sensor readings.
- Save the Obsidian note with [[Mean]], [[Median]], [[Standard Deviation]], [[Outlier]], [[Moving Average]], [[Sensor Log]], and [[Raw Data]] backlinks.
- Try interpolation only after the raw readings and noise width make sense.
Practice path
- Near-Copy Rebuild: Recreate one example, decision path, or worked explanation from Basic Statistics for Sensor Logs. Keep most givens the same, then implement, test, and explain while naming each cue you used. Use the lesson's example block when it helps.
- One-Change Transfer: Change exactly one condition, number, input, symptom, material, or constraint from the near-copy case. Then implement, test, and explain again and explain what changed.
- Mixed Review Set: Interleave this topic with one prerequisite or adjacent idea. Write three short prompts: one recall, one application, and one comparison.
- Find And Fix The Error: Invent a plausible wrong answer, unsafe step, invalid assumption, or bad classification. Mark the first point where it goes wrong, then correct it using the lesson's check.
Flashcard preview
What should stay attached to every reading?
The unit, setup, time or sample order, and what changed during the measurement.
Why keep raw readings?
Raw values let you inspect spikes, noise, and smoother side effects later.
What does standard deviation help describe?
How widely readings vary around the typical value.
When is a moving average risky?
When the project needs fast event detection or when smoothing hides the raw behavior.
What does the 'Name the measurement' step prove?
A number is not useful until the unit, sensor, sample rate, and setup are named. Check: Your note says what was measured and what changed during the sample.
What does the 'Find the typical value' step prove?
Mean or median gives a center to compare against new readings. Check: You can say whether the value is typical or unusual for this setup.
Downloadable study pack
Export the same lesson as a plain Markdown note or Anki-compatible TSV. Commands and code blocks stay plain so they work in local notes.
Related paths
Study pack check passed. Notes, cards, examples, and practice tasks are meant to keep the lesson useful outside the page.
Connected routes
Use these links like a project map: what helps before this, what this unlocks, and where it fits.
Helpful before this
Project context
What this unlocks
- Use this before deciding whether an RSSI direction peak is stable.
- Use it before filtering ESP32 or Arduino sensor readings.
- Save the Obsidian note with [[Mean]], [[Median]], [[Standard Deviation]], [[Outlier]], [[Moving Average]], [[Sensor Log]], and [[Raw Data]] backlinks.
- Try interpolation only after the raw readings and noise width make sense.
Related pages
Text lesson and video notes
This page works as a text lesson first. If you later watch a matching tutorial, use the notes pattern here to capture the build decision, timestamps, warnings, and the next practical task instead of saving a raw link.
Read the text lesson
Use the steps, examples, traps, and practice task on this page to understand the next move in a maker project.
Attach a video note
Save useful workshop or tutorial videos into an Obsidian note with timestamps, source links, and what each segment proves. The site does not need the video to be useful.
Review and practice
Download the cards, then finish the practice task before adding more links to your project notebook.
Suggest a better source video
If another tutorial explains this topic more clearly, send the title and YouTube URL. Suggestions should help the ladder, not replace it.
Topic: Basic Statistics for Sensor Logs
Continue learning this topic
Use this page as part of a project path, not as a one-off article. Save the note, review the cards, try the practice task, then choose the next lesson based on what your project exposes.
Study assets
Project context
- Build an Obsidian project notebook
- Browse Applied Data
- Next ladder clue: Use this before deciding whether an RSSI direction peak is stable.
Related references
Buy me a cup of coffee
TopicLadder is free to read. Coffee support helps turn rough maker ladders into clearer project paths, notes, cards, and practice labs.
