Heart Failure Prediction Data Analysis

The dataset for this project is from Kaggle

Kaggle dataset

Notebook Content:

• Step 1: Install and import necessary libraries
• Step 2: Reading and exploring data
• Step 3: Data cleaning and preprocessing
• Step 4: Descriptive Statistics
• Step 5: Data Analysis and visualization
• Step 6: Predictive models

Libraries needed:

• pandas
• matplotlib
• seaborn
• plotly.express and plotly.graph_objects
• numpy
• scipy.stats
• sklearn - linear regression, logistic regression, train_test_split, Decision tree classfier.

Step 1: Install and import necessary libraries

We need to install / import the necessary libraries needed for out data analysis.

Step 2: Reading and exploring data

1. Then we will read the csv file downloaded from Kaggle and read it into jupyter notebook using pd.read_csv
   The attributes of the data set are as follows:
   • Age: age of the patient [years]
   • Sex: sex of the patient [M: Male, F: Female]
   • ChestPainType: chest pain type [TA: Typical Angina, ATA: Atypical Angina, NAP: Non-Anginal Pain, ASY: Asymptomatic]
   • RestingBP: resting blood pressure [mm Hg]
   • Cholesterol: serum cholesterol [mm/dl]
   • FastingBS: fasting blood sugar [1: if FastingBS > 120 mg/dl, 0: otherwise]
   • RestingECG: resting electrocardiogram results [Normal: Normal, ST: having ST-T wave abnormality (T wave inversions and/or ST elevation or depression of > 0.05 mV), LVH: showing probable or definite left ventricular hypertrophy by Estes' criteria]
   • MaxHR: maximum heart rate achieved [Numeric value between 60 and 202]
   • ExerciseAngina: exercise-induced angina [Y: Yes, N: No]
   • Oldpeak: oldpeak = ST [Numeric value measured in depression]
   • ST_Slope: the slope of the peak exercise ST segment [Up: upsloping, Flat: flat, Down: downsloping]
   • HeartDisease: output class [1: heart disease, 0: Normal]
2. Next we will look at the shape and information of our dataset.
   • Our dataset consists of 918 rows and 12 columns.
   • Here is what our data info looks like:

Step 3: Data cleaning and preprocessing

1. Moving on, we will check for out null values and duplicated values.
2. I found some inconsistencies in our data for serum_cholesterol feature which had the value of 0 for 172 rows, which isn't possible since there can't be a 0 for serum cholesterol.
3. This is the same case for RestingBP but it only has 1 row with a 0 value.
4. Since serum cholesterol is a significant to our analysis, we can drop the rows with 0 values. Same with RestingBP.

Step 4: Descriptive Statistics

Now that our dataset is clean, we will get into some statistical analysis.

1. Here is the summary statistics of our data set:
   

   • From the above image we can note deduce that the average age of our dataset is 53 with a median of 54 and a standard deviation of 9.5. We can also see that the youngest person from our dataset is 28 and the oldest person is 77.
   • Similarly we can look at other features of our dataset and note their statistical significance.

2. Now lets look at the correlation between our features:
   

   • The highest correlation of our target variable is with Oldpeak and the lowest correlation is with MaxHr which is actually a negative value.
   • Cholesterol and Age have the next highest correlation with target variable.

Step 5: Data Analysis and visualization

Now lets look at some visualizations from our dataset.

1. I first started by diving my features into categorial and numerical, then I plotted them into count plots and distribution plots.

2. Here is a distribution plot for Age
   

   • Here we see a right skewed distribution.
   • The majority of people from our dataset lies in age group between 60-70.

3. Lets look at a distribution plot for OldPeak
   

   • Old Peak is a sign of damage to the heart muscle, the higher the value, the more damage to the heart muscle.
   • For our dataset, the peak at 0 indicates that more than 300 do not have heart damage and around 80 have moderate damage with the value of 2.
   • we can use this to identify those who are at risk for heart disease.

4. Lets look at serum cholesterol distribution.

   • We all are aware that according to research high cholesterol is a risk factor for heart diseases.
   • The normal range for serum cholesterol is less than 200 mg/dl, borderline high is between 200 to 239 mg/dl and high is above 240 mg/dl
     Source
   • In our distribution plot, the peak is at around 200 mg/dl which indicates a normal range.
   • Those who have a serum cholesterol higher than 200 mg/dl which from our distribution plot is a small number of people, they are at risk for heart disease.

5. Lets look at how many people in our dataset have heart diseases and how many don't.
   

   • 52.28% don't have any heart disease and 47.72% have heart diseases.

6. Further lets look at gender distribution between people with heart disease and those without.
   

   • As we can see that the number of male is higher than that of female hence men are more at greater risk of heart disease.

7. Lastly lets look at some box plots for serum cholesterol and old peak.
   

   • In the first boxplot for cholesterol people with no heart disease have a lower median than those with heart disease but its evident that females with heart disease have the highest median.
   • As for the old peak, people with heart disease have a higher variability than those with no heart disease. Females have a lower median than males.

Conclusion

After analyzing our data we can conclude the following

1. Males have a higher risk for heart diseases.
2. Cholesterol is a risk factor, with a correlation coefficient of 0.1.
3. Old peak is also high in correlation, with a correlation coefficient of 0.5