Datasets:

nadiCR7
/

Stroke_Prediction_Dataset

id int64	gender string	age float64	hypertension int64	heart_disease int64	ever_married string	work_type string	Residence_type string	avg_glucose_level float64	bmi float64	smoking_status string	stroke int64
9,046	Male	67	0	1	Yes	Private	Urban	228.69	36.6	formerly smoked	1
51,676	Female	61	0	0	Yes	Self-employed	Rural	202.21	null	never smoked	1
31,112	Male	80	0	1	Yes	Private	Rural	105.92	32.5	never smoked	1
60,182	Female	49	0	0	Yes	Private	Urban	171.23	34.4	smokes	1
1,665	Female	79	1	0	Yes	Self-employed	Rural	174.12	24	never smoked	1
56,669	Male	81	0	0	Yes	Private	Urban	186.21	29	formerly smoked	1
53,882	Male	74	1	1	Yes	Private	Rural	70.09	27.4	never smoked	1
10,434	Female	69	0	0	No	Private	Urban	94.39	22.8	never smoked	1
27,419	Female	59	0	0	Yes	Private	Rural	76.15	null	Unknown	1
60,491	Female	78	0	0	Yes	Private	Urban	58.57	24.2	Unknown	1
12,109	Female	81	1	0	Yes	Private	Rural	80.43	29.7	never smoked	1
12,095	Female	61	0	1	Yes	Govt_job	Rural	120.46	36.8	smokes	1
12,175	Female	54	0	0	Yes	Private	Urban	104.51	27.3	smokes	1
8,213	Male	78	0	1	Yes	Private	Urban	219.84	null	Unknown	1
5,317	Female	79	0	1	Yes	Private	Urban	214.09	28.2	never smoked	1
58,202	Female	50	1	0	Yes	Self-employed	Rural	167.41	30.9	never smoked	1
56,112	Male	64	0	1	Yes	Private	Urban	191.61	37.5	smokes	1
34,120	Male	75	1	0	Yes	Private	Urban	221.29	25.8	smokes	1
27,458	Female	60	0	0	No	Private	Urban	89.22	37.8	never smoked	1
25,226	Male	57	0	1	No	Govt_job	Urban	217.08	null	Unknown	1
70,630	Female	71	0	0	Yes	Govt_job	Rural	193.94	22.4	smokes	1
13,861	Female	52	1	0	Yes	Self-employed	Urban	233.29	48.9	never smoked	1
68,794	Female	79	0	0	Yes	Self-employed	Urban	228.7	26.6	never smoked	1
64,778	Male	82	0	1	Yes	Private	Rural	208.3	32.5	Unknown	1
4,219	Male	71	0	0	Yes	Private	Urban	102.87	27.2	formerly smoked	1
70,822	Male	80	0	0	Yes	Self-employed	Rural	104.12	23.5	never smoked	1
38,047	Female	65	0	0	Yes	Private	Rural	100.98	28.2	formerly smoked	1
61,843	Male	58	0	0	Yes	Private	Rural	189.84	null	Unknown	1
54,827	Male	69	0	1	Yes	Self-employed	Urban	195.23	28.3	smokes	1
69,160	Male	59	0	0	Yes	Private	Rural	211.78	null	formerly smoked	1
43,717	Male	57	1	0	Yes	Private	Urban	212.08	44.2	smokes	1
33,879	Male	42	0	0	Yes	Private	Rural	83.41	25.4	Unknown	1
39,373	Female	82	1	0	Yes	Self-employed	Urban	196.92	22.2	never smoked	1
54,401	Male	80	0	1	Yes	Self-employed	Urban	252.72	30.5	formerly smoked	1
14,248	Male	48	0	0	No	Govt_job	Urban	84.2	29.7	never smoked	1
712	Female	82	1	1	No	Private	Rural	84.03	26.5	formerly smoked	1
47,269	Male	74	0	0	Yes	Private	Rural	219.72	33.7	formerly smoked	1
24,977	Female	72	1	0	Yes	Private	Rural	74.63	23.1	formerly smoked	1
47,306	Male	58	0	0	No	Private	Rural	92.62	32	Unknown	1
62,602	Female	49	0	0	Yes	Private	Urban	60.91	29.9	never smoked	1
4,651	Male	78	0	0	Yes	Private	Rural	78.03	23.9	formerly smoked	1
1,261	Male	54	0	0	Yes	Private	Urban	71.22	28.5	never smoked	1
61,960	Male	82	0	1	Yes	Private	Urban	144.9	26.4	smokes	1
1,845	Female	63	0	0	Yes	Private	Urban	90.9	null	formerly smoked	1
7,937	Male	60	1	0	Yes	Govt_job	Urban	213.03	20.2	smokes	1
19,824	Male	76	1	0	Yes	Private	Rural	243.58	33.6	never smoked	1
37,937	Female	75	0	1	No	Self-employed	Urban	109.78	null	Unknown	1
47,472	Female	58	0	0	Yes	Private	Urban	107.26	38.6	formerly smoked	1
35,626	Male	81	0	0	Yes	Self-employed	Urban	99.33	33.7	never smoked	1
36,338	Female	39	1	0	Yes	Private	Rural	58.09	39.2	smokes	1
18,587	Female	76	0	0	No	Private	Urban	89.96	null	Unknown	1
15,102	Male	78	1	0	Yes	Private	Urban	75.32	null	formerly smoked	1
59,190	Female	79	0	1	Yes	Private	Rural	127.29	27.7	never smoked	1
47,167	Female	77	1	0	Yes	Self-employed	Urban	124.13	31.4	never smoked	1
8,752	Female	63	0	0	Yes	Govt_job	Urban	197.54	null	never smoked	1
25,831	Male	63	0	1	Yes	Private	Rural	196.71	36.5	formerly smoked	1
38,829	Female	82	0	0	Yes	Private	Rural	59.32	33.2	never smoked	1
66,400	Male	78	0	0	Yes	Private	Urban	237.75	null	formerly smoked	1
58,631	Male	73	1	0	Yes	Self-employed	Urban	194.99	32.8	never smoked	1
5,111	Female	54	1	0	Yes	Govt_job	Urban	180.93	27.7	never smoked	1
10,710	Female	56	0	0	Yes	Private	Urban	185.17	40.4	formerly smoked	1
55,927	Female	80	1	0	Yes	Private	Rural	74.9	22.2	never smoked	1
65,842	Female	67	1	0	Yes	Self-employed	Rural	61.94	25.3	smokes	1
19,557	Female	45	0	0	Yes	Private	Rural	93.72	30.2	formerly smoked	1
7,356	Male	75	0	0	Yes	Private	Urban	104.72	null	Unknown	1
17,013	Male	78	1	0	No	Private	Urban	113.01	24	never smoked	1
17,004	Female	70	0	0	Yes	Private	Urban	221.58	47.5	never smoked	1
72,366	Male	76	0	0	Yes	Private	Urban	104.47	20.3	Unknown	1
6,118	Male	59	0	0	Yes	Private	Urban	86.23	30	formerly smoked	1
7,371	Female	80	1	0	Yes	Self-employed	Rural	72.67	28.9	never smoked	1
70,676	Female	76	0	0	Yes	Govt_job	Rural	62.57	null	formerly smoked	1
2,326	Female	67	1	0	Yes	Private	Rural	179.12	28.1	formerly smoked	1
27,169	Female	66	1	0	Yes	Govt_job	Rural	116.55	31.1	formerly smoked	1
50,784	Male	63	0	0	Yes	Private	Rural	228.56	27.4	never smoked	1
19,773	Female	52	0	0	Yes	Private	Rural	96.59	26.4	never smoked	1
66,159	Female	80	0	1	Yes	Self-employed	Rural	66.72	21.7	formerly smoked	1
36,236	Male	80	1	0	Yes	Private	Urban	240.09	27	never smoked	1
71,673	Female	79	0	0	Yes	Private	Urban	110.85	24.1	formerly smoked	1
45,805	Female	51	0	0	Yes	Private	Urban	165.31	null	never smoked	1
42,117	Male	43	0	0	Yes	Self-employed	Urban	143.43	45.9	Unknown	1
57,419	Male	59	0	0	Yes	Private	Rural	96.16	44.1	Unknown	1
26,015	Female	66	0	0	Yes	Self-employed	Urban	101.45	null	Unknown	1
26,727	Female	79	0	0	No	Private	Rural	88.92	22.9	never smoked	1
66,638	Female	68	1	0	No	Self-employed	Urban	79.79	29.7	never smoked	1
70,042	Male	58	0	0	Yes	Private	Urban	71.2	null	Unknown	1
32,399	Male	54	0	0	Yes	Private	Rural	96.97	29.1	smokes	1
3,253	Male	61	0	1	Yes	Private	Rural	111.81	27.3	smokes	1
71,796	Female	70	0	1	Yes	Private	Rural	59.35	32.3	formerly smoked	1
14,499	Male	47	0	0	Yes	Private	Urban	86.94	41.1	formerly smoked	1
49,130	Male	74	0	0	Yes	Private	Urban	98.55	25.6	Unknown	1
28,291	Female	79	0	1	Yes	Private	Urban	226.98	29.8	never smoked	1
51,169	Male	81	0	0	Yes	Private	Urban	72.81	26.3	never smoked	1
66,315	Female	57	0	0	No	Self-employed	Urban	68.02	37.5	never smoked	1
37,726	Female	80	1	0	Yes	Self-employed	Urban	68.56	26.2	Unknown	1
54,385	Male	45	0	0	Yes	Private	Rural	64.14	29.4	never smoked	1
2,458	Female	78	0	0	Yes	Private	Rural	235.63	32.3	never smoked	1
35,512	Female	70	0	0	Yes	Self-employed	Rural	76.34	24.4	formerly smoked	1
56,841	Male	58	0	1	Yes	Private	Rural	240.59	31.4	smokes	1
8,154	Male	57	1	0	Yes	Govt_job	Urban	78.92	27.7	formerly smoked	1
4,639	Female	69	0	0	Yes	Govt_job	Urban	82.81	28	never smoked	1

End of preview. Expand in Data Studio

Stroke Prediction Dataset Analysis

Presentation Video

Project Overview

The goal of this project is to predict the likelihood of a patient suffering a stroke based on demographic, health, and lifestyle parameters. Stroke is a leading cause of death and long-term disability worldwide, and early identification of high-risk individuals can significantly improve prevention strategies and clinical outcomes.

Dataset Summary

Source: Kaggle – Stroke Prediction Dataset by fedesoriano
Size: 5,110 rows × 12 features
Target Variable:
- stroke → (1 = patient had a stroke, 0 = no stroke)

Key Feature Categories

numeric_columns- age - bmi - avg glucose level- stroke
categorical_columns- gender - hypertension - heart disease - ever married - work type - Residence type - smoking status

Exploratory Data Analysis & Insights

1. Data Cleaning & Handling

'id' coulm:

The id column was dropped from the dataset.
- Reason: The id column serves only as a unique identifier for each record and does not contain any meaningful information related to stroke prediction. Including it in the analysis or model could introduce unnecessary noise without improving predictive performance.

Missing Values

The bmi column contains 201 missing values.
- These null entries were addressed using Mean Imputation. Each missing value was filled with the average BMI calculated from the rest of the column to maintain the overall distribution of the dataset without losing significant rows of data.

Outlier Detection

BMI:

Action: Outliers were identified but retained.
Reason: High BMI values are clinically significant and strongly associated with increased health risks, including stroke. Removing these records would result in the loss of important information about how extreme BMI values influence the target variable.
Glucose Level:

Action: Outliers were identified but retained.
Reason: Extreme glucose levels are medically relevant indicators, particularly in the context of stroke risk. Retaining these values ensures that the model captures the full spectrum of possible health conditions.
Gender Variable:

Action:A third category, "Other," was identified in the data and removed.
Reason:The count for this category was extremely low(only 1). Removing it simplifies the analysis without impacting the overall ability to understand the target value, given the abundance of data for "Male" and "Female."

Data Observations & Limitations

Smoking Status:

A large number of records are categorized as 'Unknown'.
- Note: Because of this high frequency of missing information within the category, any conclusions drawn regarding the impact of smoking status on stroke risk should be taken with a grain of salt.

2. visualizations

-percentage of patients compared to the number of percentage of healthy people:

In the dataset of the stroke patients, the distribution was as follows:
- Healthy (no stroke): 95.1%
- Patients (stroke): 4.9%

-Correlation matrix for numerical features: -The goal is to find interesting relationships—either between the independent variables themselves or between them and the target variable

Relationships between independent variables themselves-here is a notable correlation between age and bmi (0.33), as well as age and glucose level (0.24).
- Relationships between independent variables and the target variable: We will talk about the topic in the research question

-Initial exploration of variable distributions:

Age: There is a presence of people of all ages, with a peak around the 50-60 age range. implies a diverse data set.
BMI: Most of the data is concentrated between 20 and 40.There is a "long tail" extending toward the right, indicating a few individuals with very high BMI values
Avg Glucose level: The initial, much larger peak is around 80-100, which represents the "normal" range for most people. A secondary, smaller peak appears around 200-230, presumably representing a subset of the "unhealthy" population.
Stroke: The vast majority of the data set did not have a stroke. The small bump on the right side shows that a minority had a stroke.

3. Research Questions:

Is there a correlation between age and the prevalence of stroke?
Do BMI and average glucose levels serve as significant predictors for stroke?
Does smoking status increase the likelihood of experiencing a stroke?
Are people with heart disease and Hypertension More likely to have a stroke??

4. Answers:

Is there a correlation between age and the prevalence of stroke? Yes! - - -The data show a clear age difference between the groups: people who have experienced a stroke are significantly older compared to those who have not.
Do BMI and average glucose levels serve as significant predictors for stroke? Partial answer - - - While glucose levels show a clear correlation to stroke, BMI shows a slight association, the data is inconclusive. Therefore, We cannot provide a definitive answer.
Does smoking status increase the likelihood of experiencing a stroke? Partial answer - - It is difficult to say whether smoking causes a stroke. Because we have already mentioned that a large number of people when asked about smoking gave the following answer - "unknown". Therefore, any result of this sample, we took it with a grain of salt.
Are people with heart disease and Hypertension More likely to have a stroke? Yes! - - -The data show that hypertension and heart disease are a common cause of stroke.

5. Insights:

1. Age appears to be the main determinant of stroke risk, with older people being more likely to experience a stroke.
2. Heart disease appears as a major factor, strongly associated with a higher likelihood of stroke.
3. The presence of Hypertension is also an important risk indicator, increasing the probability of a stroke occurring.
4. High glucose levels may contribute to stroke risk, suggesting a possible link between impaired blood sugar regulation and stroke.
5. A higher BMI shows a slight association with stroke, the data is inconclusive. Therefore, We cannot provide a definitive answer.
6. No clear connection was found between smoking and stroke in this dataset. That said, with nearly 30% of the values missing, the data may not be consistent enough to draw a firm conclusion.

Overall:

Only 5% of the individuals in this dataset suffered a stroke. Due to this significant class imbalance, the insights may not represent the general population perfectly. All analyses and conclusions are strictly based on the patterns found within this specific dataset. The risk of stroke is mainly driven by a person's age and physical health. Age stands out as the most significant factor, with the risk increasing sharply as the patients get older. This is closely followed by hypertension, heart disease, and high glucose levels, all three of which act as strong predictors of stroke occurrence in this data set. It is also important to note that for some variables, such as smoking and BMI, the high amount of missing data made it difficult to draw a firm conclusion. Ultimately, the data indicate that stroke is a combined condition, where a combination of aging and poor physiological condition creates the highest level of risk.

Author

Nadav Benaiah

Downloads last month: 106