zconf_interval_two_samples

create function 
zconf_interval_two_samples
(total_a float, conv_a float,  total_b float, conv_b float, alpha float)

--INPUTS:
-- total_a = total number that saw A
-- conv_a = converters for version A
-- total_b = total number that saw B
-- conv_b = converters for version B
-- alpha = alpha value (0.05?)

RETURNS float
--OUTPUT: 
--Returns confidence interval
STABLE 
AS $$

    from scipy.stats import norm  
    import numpy as np  
    
    p1 = conv_a/total_a
    p2 = conv_a/total_b  
    se = np.sqrt(p1*(1-p1)/total_a + p2*(1-p2)/total_b)  
    z_critical = norm.ppf(1-0.5*alpha)
    return p2-p1+z_critical*se

$$LANGUAGE plpythonu;