Driver's License Type Prediction

Introduction

This project aims to classify the license type of the drivers located in Dubai.

Dataset

We have Data from Driver's population census in Dubai. The dataset contains overall 5 columns of which 1 column contain content in Arabic linguistics and the rest are numeric and object types. The dataset is provided in the repository.

Dataset Statistics:

Here is sample statistics of the data.

• Name: Drivers_Population_Census
• Mode: Tabular Data
• Number of Samples:
  • Train: 2,247
  • Test: 561
• Type: Classification
• Number of Classes: 4
• Classes Name:
  • Registration
  • Transfer of Ownership
  • Change number plate
  • Update Info

Pre-processing

The preprocessing steps of the proposed project are the following:

1. Made new list and replaced Arabic Linguistics content into English linguistics
2. Apply One-hot encoding on Gender and age_group columns
3. Apply Label encoding on Nationality and License_type columns
4. To avoid over sampling we use sampling_strategy method from SMOTE module to avoid class imbalancing

Models

For the Traffic License Prediction in Dubai following models were trained with the tuned hyperparameters:

• K-Nearest Neibhour

  • n_neighbors=3

• Decision Tree

  • GridSearchCV with parameter
    • criterion : ['gini', 'entropy'],
    • splitter : ['best', 'random'],
    • max_depth : [None, 10, 20, 30, 40, 50],
    • min_samples_split : [2, 5, 10],
    • min_samples_leaf : [1, 2, 4],
    • max_features : ['auto', 'sqrt', 'log2', None]
  • cross validation = 5
  • scoring = f1
  • estimator=dt_model

  Random Forest

  • GridSearchCV with parameter
    • criterion : ['gini', 'entropy'],
    • min_sample_leaf = [1, 2, 4]
    • max_depth : [None, 10, 30],
  • cross validation = 5
  • scoring = accuracy
  • estimator=rf_model

Results

Metrics	KNN	Decision Tree	Random Forest
Accuracy	0.796889	0.829826	0.852242
Precision	0.796594	0.830111	0.852526
Recall	0.796921	0.829830	0.852233
F1-Score	0.792896	0.829060	0.851289

Author

Arslan Ahmad
Research Officer
National Center of Artificial Intelligence (NCAI)
AL-Khawarizmi Institute of Computer Science (KICS) UET Lahore