We have a very simple use case, when we take some data X, y and fir it to LinearRegression() like below. and try to predict some more value.
We have set the seed to some non-zero number again mentioned below.
np.random.seed(42)
regr = LinearRegression()
regr.fit(x, y)
pred = regr.predict(pred_set)[0]
The above prediction where as always same in one machine, changes when ran from different machine, specifically when we ran it in some CICD test pipeline on cloud environment.
As per my understanding, setting seed to a specific number should confirm the reproducibility, but is there anything I am missing here. Or If I need to set seed in some other way.
Kindly, let me know.