Week 7 — Career Path Predictor

Hello everyone again from a good week! Although this week has been a bit tiring for us, our project has started to take shape. Let’s see the developments together..

We started this week with the feedback we received as a result of TA meetings. Upon the transfer learning proposal received here, we started to learn the transfer learning concept. Then we looked at the studies and datasets suitable for our own project in order to implement this. Unfortunately, we did not find many examples for tabular data. We made a transfer learning application using a dataset that we saw as the most suitable, but unfortunately we had to abandon this idea due to some inconsistency.

While trying to improve our Neural Network model (Because we faced an overfitting problem and we are trying to solve it.), we also tried different algorithms. We created new models using XGBoost, AdaBoostClassifier, GaussianNB, GradientBoostingClassifier, KNeighborsClassifier algorithms and analyzed the results. You can see the Log Loss graph created as a result of the XGBoost model.

However, as a result of these examinations, we saw that the Logistic Regression model has the highest accuracy. Then we went to the fine tuning stage and tried to find the best parameters for logistic regression with GridSearchCV.

We ended the week by planning the rest of this stage. See you next week! Stay well and AI…

- Melike Nur Dulkadir

- Sare Naz Ersoy