SQL Data
	feature	category	stroke_status	count								Expected Values			Chi-Square Test
	ever_married	married	had stroke	81		Sum of count	Column Labels					had stroke	no stroke		feature	p-value	significant
	ever_married	married	no stroke	2333		Row Labels	had stroke	no stroke	Grand Total		married	53.22390985	2360.77609		ever_married	1.7144E-09	yes
	ever_married	never married	had stroke	9		ever_married	90	3992	4082		never married	36.77609015	1631.22391		gender	0.3238055	no
	ever_married	never married	no stroke	1659		married	81	2333	2414						heart_disease	1.93107E-14	yes
	gender	female	had stroke	48		never married	9	1659	1668		female	52.56246938	2331.437531		hypertension	1.40644E-05	yes
	gender	female	no stroke	2336		gender	90	3992	4082		male	37.43753062	1660.562469		residence_type	0.569317433	no
	gender	male	had stroke	42		female	48	2336	2384						smoking_status	0.000122057	yes
	gender	male	no stroke	1656		male	42	1656	1698		heart disease	2.116609505	93.88339049		work_type	0.004206256	yes
	heart_disease	heart disease	had stroke	13		heart_disease	90	3992	4082		no heart disease	87.88339049	3898.11661
	heart_disease	heart disease	no stroke	83		heart disease	13	83	96						The Chi-Square Test indicates that for patients under 65, gender and residence_type are not statistically significant predictors of stroke, as their p-values exceed the 0.05 significance threshold. Gender and residence_type features will be dropped from the analysis, moving forward.
	heart_disease	no heart disease	had stroke	77		no heart disease	77	3909	3986		hypertension	5.908868202	262.0911318
	heart_disease	no heart disease	no stroke	3909		hypertension	90	3992	4082		no hypertension	84.0911318	3729.908868
	hypertension	hypertension	had stroke	16		hypertension	16	252	268
	hypertension	hypertension	no stroke	252		no hypertension	74	3740	3814		rural	44.66927976	1981.33072
	hypertension	no hypertension	had stroke	74		residence_type	90	3992	4082		urban	45.33072024	2010.66928
	hypertension	no hypertension	no stroke	3740		rural	42	1984	2026
	residence_type	rural	had stroke	42		urban	48	2008	2056		formerly smoked	12.89808917	572.1019108
	residence_type	rural	no stroke	1984		smoking_status	90	3992	4082		never smoked	32.74130328	1452.258697
	residence_type	urban	had stroke	48		formerly smoked	22	563	585		smokes	14.52964233	644.4703577		Mann-Whitney Test
	residence_type	urban	no stroke	2008		never smoked	24	1461	1485		unknown	29.83096521	1323.169035		feature	p-value
	smoking_status	formerly smoked	had stroke	22		smokes	25	634	659						bmi	1.06E-06
	smoking_status	formerly smoked	no stroke	563		unknown	19	1334	1353		children	15.14698677	671.8530132		avg_glucose_level	0.0001
	smoking_status	never smoked	had stroke	24		work_type	90	3992	4082		govt_job	11.59725625	514.4027438
	smoking_status	never smoked	no stroke	1461		children	2	685	687		never_worked	0.485056345	21.51494366		The Mann–Whitney U Test was conducted using Python, as Excel does not support this test natively. The script can be found in the python_stat_test folder under the filename mann_whitney_test.py. Both BMI and average glucose level returned p-values significantly below the 0.05 threshold, indicating that they are statistically significant predictors of stroke in patients under 65.
	smoking_status	smokes	had stroke	25		govt_job	16	510	526		private	53.13571779	2356.864282
	smoking_status	smokes	no stroke	634		never_worked		22	22		self-employed	9.634982852	427.3650171
	smoking_status	unknown	had stroke	19		private	59	2351	2410
	smoking_status	unknown	no stroke	1334		self-employed	13	424	437
	work_type	children	had stroke	2		Grand Total	630	27944	28574
	work_type	children	no stroke	685
	work_type	govt_job	had stroke	16
	work_type	govt_job	no stroke	510
	work_type	never_worked	no stroke	22
	work_type	private	had stroke	59
	work_type	private	no stroke	2351
	work_type	self-employed	had stroke	13
	work_type	self-employed	no stroke	424