Week 5 — Career Path Predictor
Hello everyone! There have been surprising developments this week in our project, which we are looking forward to completing in order to learn about our career path as soon as possible. Without wasting time, let’s examine this week and the changes in our project together.
We started this week by continuing to review articles containing studies similar to our project. As a result of these investigations, we came across a new dataset, which we think is a version of the dataset we use with more features and observations. In order to get rid of the surprise we were faced with this new development, we tried our new dataset in our studies and observed the accuracy result it gives in a decision tree model. As a result, the accuracy we achieved was 17.63%.
We consulted with Associate Professor Erkut Erdem about which dataset we should proceed with. After the feedback we received, a new idea came to our minds, for now we are keeping it as a surprise to talk about next week, but we would like to state that we are very excited about this. Using this model we created, we examined the importance of each feature in the dataset in the model. Based on our reviews, we started our Feature Engineering studies. We have derived 5 new features and added them to our dataset. Of course, we did not forget to apply one hot encoding to the necessary ones! Depending on our creativity, we can improve this part in the future, add different columns and maybe drop some columns that will make the model better.
To see how much our work has improved our model, we created a decision tree model again and we saw that the accuracy value reached 17.96%. Then we created LogisticRegression, Support Vector Machine, Random Forest Classifier models to examine the success of our dataset in different models. Among them, Logistic Regression achieved the best result with 20.78% accuracy.
Thus, we have left behind another week that increased our excitement for the future of our project, in which we passed the critical stages. See you next week. Take yourself with AI!
-Melike Nur Dulkadir
-Sare Naz Ersoy