I am fairly new to PyMC3 and I am trying to model the uncertainty propagation during the calibration process of a sensor. As such, the outputs of my model are [Y] and the inputs are [X]. xdata and ydata are the np.arrays containing my calibration data. My model looks` like this: def volt_est(w,x): …

Hi Santiago- Looks to me like you’re calibrating a model in terms of P(Y|X) but then want to sample from P(X|Y). This is doable since the joint distribution P(X,Y) is what’s modeled, but sampling from a pre-image is not as straightforward as posterior prediction. I can think of two ways to do this. …

Making predictions on model inputs based on observed outputs

Questions

chartl March 18, 2021, 2:29am 4

I would also recommend using a root finder (like sp.optimize.newton) since it’s much faster; the use of a least-squares minimization was just an illustrative hack.

1 Like

Topic		Replies	Views
The pymc3 way: Linear regression and inferring given posterior/trace Questions	4	1780	May 9, 2018
Finding posterior for calibration using saved Gaussian model in pymc3 Questions theano	2	1352	November 15, 2017
Odd results in model prediction using pymc.sample_posterior_predictive v5 linear_model , modeling	9	1257	September 24, 2022
Use saved gaussian model from sci-kit in pymc3? Questions theano	3	1573	November 30, 2017
How to: calibration -> inverse prediction	5	275	December 20, 2023

Making predictions on model inputs based on observed outputs

Related topics