Let's start with some important terms you'll come across when calibrating your devices.

By the end of this lesson, you’ll be able to understand:

• Definitions of important concepts related to calibration

Correlation or R²: The Pearson correlation coefficient, or R² , is a measure of how well correlated two variables are to one another. Correlation tells you whether two variables change together at the same rate. R² can range between 0 (not correlated) and 1 (perfectly correlated). You can read more about R² in the Knowledge Base article here.

Error or RMSE: Error is the amount that a variable differs from the expected value, and Root Mean Square Error (RMSE) is one commonly-used measure of error. It can range from zero to infinity, and its units are the same as the two variables being compared. You can read more about RMSE in the Knowledge Base article here.

Calibration model: A calibration model is the general formula or algorithm that takes sensor readings and returns the calibrated measurement. It is fitted, or “trained” on, collocation data to develop a calibration profile.

Calibration: is the process of evaluating and adjusting the Clarity Node-S measurements to ensure that they are reporting accurate data. Calibration improves the accuracy and reliability of your data. Clarity’s rigorous, patent-pending Remote Calibration process ensures your air quality measurements are scientifically validated and defensible. 

Global Calibration: These calibrations are the result of rigorous development involving the collocation of Clarity devices with reference monitors at numerous global sites, covering a broad spectrum of pollution and environmental conditions. These calibrations significantly boost accuracy compared to raw sensor data. They are designed to work well in a broad range of environments but are less individually tailored to correct in any one specific environment. All Clarity Node-S devices come equipped with global calibration applied to PM_2.5 and NO₂.  

Custom Collocation-based Calibration: During a collocation-based calibration, a calibration model is developed using data from a reference monitor. That calibration can then be applied to data from the low-cost sensor to adjust the data so that it is more accurate.  While all Clarity Node-S devices come equipped with global calibration PM_2.5 and NO₂ applied, for best results, a custom collocation-based calibration is recommended as it is tailored to your project's specific environmental conditions. 

Performance targets: Performance targets are statistical thresholds that are used to evaluate whether a sensor’s data is of sufficiently high quality. These targets are often developed by regulators or government agencies to account for variability in performance of low-cost sensors.

Detection limit: A detection limit is the concentration below which a monitor can no longer accurately measure a given pollutant. It is important to understand the detection limit of whatever instrumentation you’re working with as you start to analyze and interpret the data.