R² Correlation

The R² correlation tool measures how well a Sensorbee sensor's data tracks a reference dataset. It uses the coefficient of determination (R²) — the proportion of variance in the dependent variable explained by the model — and the Pearson correlation coefficient (r) for direction and strength.

Use it to validate a calibration, decide whether the sensor is fit for compliance reporting, or compare deployment sites against a reference station.

For the calibration step that follows, see Calibrate.

How to run a correlation

1. Select sensor

Pick the sensor on the device whose data you want to assess.

2. Set interval / resolution

Pick the time resolution to use — minute, hour, day, etc. — based on what makes sense for the data you're comparing.

3. Choose the Correlate action

Picking Correlate initiates the R² calculation between the sensor's values and your reference data.

4. Select reference data

Pick the reference dataset to compare against. This can be data from a different sensor, a different device, or a known standard or benchmark.

5. Select installation

Pick the installation the sensor is part of — this scopes the correlation to that site, project, or grouping.

Reading the output

The result graph shows three things:

R² value

The strength and direction of the relationship, by Pearson correlation coefficient:

Linear equation

Linear regression of the form Y = BX + A:

• ·B (the slope) = r × (Sy / Sx), where Sy and Sx are the standard deviations of y and x.
• ·A (the intercept) = mean(y) − (B × mean(x)).

Statistical significance

Computed at a significance level of 0.001:

• ·p < 0.001 — relationship is statistically significant.
• ·p > 0.001 — relationship is not statistically significant.

Next: span calibration

To apply a span calibration based on this correlation, continue to Calibrate.