This model contains 301 records concerning heart disease diagnosis. All attributes (8) are numeric-valued. The data was collected from the Cleveland Clinic Foundation. "Num" is the target value of my model. Thus, if num = 1, the patient is more than 50% likely to have any heart disease, but if num = 0, the probability of having the disease is less than 50%. For the model, I used only 8 attributes of the dataset, which originally consisted of 14 attributes. The reason for this was that the number of records in the 6 attributes of the original dataset that I did not take into account for calculation, was very small.

(? sign represents the missing value for corresponding attribute)

I deleted only one row in dataset, because that record has a missing value for “thal” attribute:

df_cleveland.drop(df_cleveland[df_cleveland['thal'] == '?'].index, inplace = True)
df_cleveland['thal'] = df_cleveland['thal'].astype(float)

70% of the data was used for training and the remaining 30% was used for testing. However, due to the large number of attributes but only a small number of training records in the LogisticRegression module, the error is varying noticeably in each iteration of the lbfgs algorithm process, so I gave the number of iterations manually:

from sklearn.linear_model import LogisticRegression
my_model = LogisticRegression(solver='lbfgs', max_iter=100000)

As can be seen from the name of the processed.cleveland.data file I selected, the irrelevant attributes in the dataset have been deleted, so there is no need to further look for irrelevant attributes among the 8 attributes to increase accuracy.

Based on the content of dataset, we can say that due to the importance of the diagnosis of heart disease, the number of False Negative cases in the confusion matrix should be as small as possible. From the results of the Confusion Matrix and Recall score (~ 80%) we can say that our model shows good results for the above criteria. In general, based on our AUC value (~ 90%), the model's ability to distinguish between two classes is acceptable.