Model for Predicting Suppliers in the Financial Sector of the Vehicle Manufacturing Company in the City of Pereira

Abstract

This research work presents the development of a model using data mining techniques to identify financial variables in a manufacturing company of automotive vehicle bodies in Pereira. The study is structured into four key phases. The first phase focuses on data preprocessing, including characterization, normalization, and dimensionality reduction using PCA, Relief, and Correlation. The second phase applies unsupervised learning with K-means and Gaussian Mixture Models (GMM) to cluster and validate data based on a defined target variable. In the third phase, supervised classifiers such as Bayesian Classifier, Artificial Neural Networks, Support Vector Machines, and KNN are employed to predict supplier efficiency, optimizing investment and costing processes. Finally, the fourth phase integrates preprocessing and prediction into a practical form, using libraries such as Plotly and Dash for detailed visualizations, and tools like GitHub and Heroku for application development. This study highlights the importance of artificial intelligence in business decision-making, demonstrating how data science techniques and visualization tools can facilitate the interpretation and utilization of data analysis results.