For a project involving modelling and control of a pharmaceutical tabletting line, I created an app which predicts the torque based on a set of 3 given inputs. I used machine learning and statistical analysis to create a linear regression model, and then MATLAB to better visualise that model.

First, a linear regression model needs to be generated - this is done through taking the data gathered experimentally, cleaning it with a data mask, and then using the normal equation to gain a single line. This line is chosen because it has the least error between all the data points.

Once the calculations are done, it's simple to take in 3 user inputs and effectively work out the matching torque value. There is an additional calculation for liquid flow rate, as the Consigma-25 uses liquid to solid ratio as the input and so that was mirrored in the app.