Getting the result from the KS test:
from scipy import stats
stats.kstest(myarray, 'gamma', (a, b))
we conclude that myarray does not follow a gamma distribution in case p<0.01 (when rejecting the null hypothesis)?
In addition, when we calculate the a and b of the gamma distribution (used for comparing our data) from our data, we cannot use the KS test results?
Yes - that's what the p value indicates - the probability of observing a KS statistic as extreme as this if my_array follows the gamma distribution is less than 1 percent.
I wouldn't say that it is meaningless, in the sense that deriving parameters and then doing a KS test on the same data is not the same as a tautology like 1 =1, but it is of not much value practically. It is best to use a different array for testing, and in case an array is not available - then you can do a parameter estimation on, say, 50 percent of the data and the KS test on the other 50 percent.
Also, this question is better suited for the stats.stackexchange.com forum
