What question I have answered:
1. If there is significant difference in gender and age factors on drug use?
2. AMITRIPTYLINE is the most effective drug among those four groups?
3. Among AMITRIPTYLINE, BUPROPION, and CITALOPRAM those three method which one is the best?
4. Which two drug use appear together the most on treatment of Depression.
Considering the different age and gender, different drugs will definitely have different effects on different groups. So I designed the first question. The whole group of dataset is divided into four groups, namely young men, young women, old men and old women, which can help doctors to find out the suitable method for each group. drug use is only one method of treating depression. If drug use is not very effective for a certain group, then drug use can be excluded at the beginning of treatment consideration. This will also save time.
In the case of helping doctors to determine the drug use that will be implemented. We will want to know the effectiveness of each drug for different groups, so we designed a second question. We conclude that AMITRIPTYLINE is the most effective method for young men. However, different drugs may be most effective for different groups. In this case, we can target different groups with different drugs. The ultimate goal is also to save time.
The third question is without considering gender and age. Which drug would be more effective for all groups. The goal of this question is to help the doctor find out which drug is more effective in the total population. Identify the most effective one. This will help the doctor to understand the best and most common methods among the many drug uses.
The fourth question is to consider the case where two drugs are used together. Realistically, it is possible that the effectiveness of using only one drug is not very high. So when considering the use of two drugs together which two combinations together would be the best?
First is for the conclusion of Naive Bayes method on the record data. This section basically explains about if there is significant difference in gender and age factors on drug use. We can see that training data set and test data set all have about the same accuracy which is about 53%. And for both training data set and test data set it all have the highest accuracy of about 88% for the second group, namely, the older male group. It was followed by young men, meaning men who is younger than 40, with an accuracy rate of about 25 percent. And the last two group young women and old women have pretty bad accuracy. Which indicating that these drugs have no significant change in women, so it can be said that the effect on women is not great.
Second is for the conclusion of SVM method and based on the accuracy above I use linear kernel to make conclusion about AMITRIPTYLINE is the most effective drug among those four groups. And the result is that it has the most significant effect on young male group. Since I only show one drug effectiveness in my website. And if you want to know more about each drug use you can download my code and simply change the drug name according to my data. And then you will get the result.
Third is for the conclusion of the Decision Tree method. This section answered about which drug among those three drugs(AMITRIPTYLINE, BUPROPION, and CITALOPRAM) are most fit the Decision Tree method which means are more effective among all group. The decision tree model is best fit for drug AMITRIPTYLINE compared to the other 2 drugs. Since AMITRIPTYLINE have the highest accuracy score which is 82.4%. So, AMITRIPTYLINE is the most affective drug among those 3 drugs.
Also, in the group old women and young men have the highest and same f1 score which is 0.84. And since we have the conclusion from SVM method that in the use of AMITRIPTYLINE has the most significant effect on young male group. So we can conclude that in the case of AMITRIPTYLINE drug use, it can perform good for young men in compare BUPROPION and CITALOPRAM.
Fourth is for the conclusion of Clustering. It did not have conclusion about drug use and only help us to cluster the dataset. And overall, the Kmeans method cluster the data point best. Since I have four groups and it also indicate there will be 4 clusters. So based on this dataset, Kmeans method perform the best.
For the conclusion of ARM method, we can conclude that CITALOPRAM and ESCITALOPRAM appear together the most on treatment of Depression.
Now I have given answer for several questions. I think it can help doctors to make some recommendations on drug use, although the drugs needed are different for different people. But after the data analysis, we can make some narrow down of the drug use, which can treat the depression more efficiently and help the patient to get rid of the depression in a shorter time. Of course, the analysis I did leaves a lot to be desired. And there are more questions that can be analyzed. If you want more in-depth analysis and discussion of this project, you can contact me through the homepage.