Fisher’s exact test is a statistical test that determines if two category variables have non-random connections or we can say it’s used to check whether two category variables have a significant relationship. In this article let’s learn how to perform Fisher’s exact test. In python fisher exact test can be performed using the fisher_exact() function from the SciPy library.
Syntax: scipy.stats.fisher_exact(table, alternative=’two-sided’)
Parameters:
table : array like 2X2 contigency table. negative integers aren’t allowed.
alternative: it’s an optional value which represents the alternative hypothesis. values can be {‘two-sided’, ‘less’, ‘greater’}, by default it is two sided
- less : one sided
- greater: one sided
Returns : oddratio and p_value.
Types of hypothesis
- null hypothesis: no associations between the rows and columns, or the classifications are independent.
- alternative hypothesis: there are associations between the rows and columns, or the classifications are dependent.
The 2X2 contingency table we use for the example is :
2 men and 8 women follow a diet. 7 men and 3 women don’t.
# importing packages import scipy.stats as stats
# creating data data = [[2, 8], [7, 3]]
# performing fishers exact test on the data odd_ratio, p_value = stats.fisher_exact(data)
print('odd ratio is : ' + str(odd_ratio))
print('p_value is : ' + str(p_value))
|
Output:
odd ratio is : 0.10714285714285714 p_value is : 0.06977851869492728
if we take the level of significance to be 0.05, we fail to reject the null hypothesis as the p_value is above 0.05. the alternative hypothesis is rejected.