For the last project of my data analysis training, I worked as a consultant for the National Organization for Counterfeit Money Control (ONCFM). The objective was to develop a Python-based model capable of identifying counterfeit euro bills by analyzing specific dimensions and features of the bills.
-
Data Preprocessing and Analysis:
-
Clean and preprocess the dataset, handling missing or erroneous values, using linear regression to impute missing data.
-
Perform exploratory data analysis with Python to identify key features for modeling and understand the dataset.
-
-
Predictive Modeling:
-
Experiment with various machine learning models in Python, including logistic regression and k-means, to build a model capable of detecting counterfeit bills.
-
Test and evaluate the models to select the most accurate algorithm for final deployment.
-
-
Model Evaluation and Tuning:
-
Fine-tune the chosen model and validate it using a test dataset to ensure robustness and accuracy.
-
Prepare the Python model for live testing, ensuring it can handle new, unseen data effectively.
-
Build an executable application which encapsulate the trained model, enabling users to easily use it without needing to intercat with Python.
-
-
Reporting and Presentation:
-
Present the complete process, including data treatments, algorithms explored, and final model selection, to the ONCFM project lead.
-
Demonstrate the model’s performance using a new dataset provided during the presentation.
-
This project allowed me to apply my Python skills to predictive modeling and data processing, successfully creating a model to detect counterfeit bills.