Exploring factors have influence on culvert early deterioration based on machine learning techniques
1. Introduction should include:
a. Importance of culvert maintenance and condition prediction.
b. Introduce and compare Decision tree and SVM.
c. introduce ANN
2. Develop a Decision Tree or SVM model to separate culvert data into 2 groups in the excel file based on GENERAL APPRAISAL. Group one with rating 0-6 and other 7-9.
3. Develop an ANN predictive model for the group with 0-6 GENERAL APPRAISAL and conduct sensitivity analysis to identify the impact from each features.
4. For all models:
Dependent variable : General appraisal
Independent variable: except for object ID, longitude, latitude and other condition ratings, all other (shape, length, pH, material, etc.) should be evaluated to find the best feature classes
5. Model validation process must be included