Calibrations: Sensor Coefficient Explorer
Ridge · L2-regularised · PM2.5 Correction
14 sensors · 168 terms
Ridge Regression. Linear model with an L2 penalty that shrinks coefficients — more stable when PM2.5, RH and Temp are correlated.
Sensor Brand
Measurement Period
Correction Equation · Ridge
PM2.5_corrected = -19.2440 + 1.5560 × PM2.5 − 0.4460 × Rel. Humidity + 0.1790 × Temperature
x1=PM2.5 · x2=RH · x3=Temp
Coefficients & Std. Deviations
positive negative
Intercept
-19.2440±0.3090
PM2.5
+1.5560±0.0090
Rel. Humidity
-0.4460±0.0090
Temperature
+0.1790±0.0240
Multiple Linear Regression. Ordinary least-squares fit of corrected PM2.5 on raw PM2.5, relative humidity and temperature.
Sensor Brand
Measurement Period
Correction Equation · MLR
PM2.5_corrected = 7.1080 + 1.5620 × PM2.5 − 0.4430 × Rel. Humidity + 0.2010 × Temperature
x1=PM2.5 · x2=RH · x3=Temp
Coefficients & Std. Deviations
positive negative
Intercept
+7.1080±3.0860
PM2.5
+1.5620±0.0090
Rel. Humidity
-0.4430±0.0200
Temperature
+0.2010±0.0690