Pycaret anomaly detection setup: ValueError: Setting a random_state has no effect since shuffle is False

Question

I have recently transitioned from R to python, and I am not sure how to problem solve the following.

When I run the setup for pycaret anomaly detection, following the instructions that can be found here, on my own data I get the following error.

# Inital setup
s = setup(data, session_id = 1230)
ValueError: Setting a random_state has no effect since shuffle is False. You should leave random_state to its default (None), or set shuffle=True.

The only difference from the example is that I have some additional xregs (base_price(float64), sale_price(float64), promotion_flag(int64; 0 or 1)). As I understand it, this shouldn't affect the results, after all there are a number of time features. Everything else is the same. So I don't understand why this error is occurring.

I tried the answers on this stackoverflow question, along with adding shuffle=True, but these all resulted in a unexpected keyword argument.

I appreciate this may be a bit basic, but the error doesn't make sense to me, based on what I have done. Thanks for your time.

Python 3.7.10

Answer 1

The answer to this question is that the environment I was using in anaconda had library versions such as numpy that were too new for pycaret to work with, for example, pycaret need numpy (1.19.5 and will not work with newer).

My solution was to create a new environment in anaconda, which used pip install pycaret[full], and added nothing else to the environment. It worked after this.