MachineLearning

Codes for the course of Machine Learning at UFRN PPGEEC2318 - Machine Learning - Prof. Ivanovich Silva

Student: Lucas Aires da Costa Silva

Activity 01

Study of Classification Problem and Data Manipulation

Notebook: Atividade01_db02.ipynb

• Selecting a dataset: "Heart-Disease-App"

• Preparation of a dataset, manipulating categorical and numerical data;

• Spliting a dataset in train, validation and test data;

• Model configuration for traning;

• Classification problem;

• Metrics (accuracy, precision and recall, true and false positive rates) usage to observe the results.

Final Project - Part I

Study of Multiclass Classification Problem (3+)

1. First step - Study_week11.ipynb:

   • Testing the algorithm by changing the number of features, to see how it effects the results of training;
   • Changing the blocks of processing to see its effects (we observed a problem, because of the image size);

2. Second step - Multiclass_pt1_Lucas_Aires.ipynb:

   • Selecting a dataset of images, for multiclass classfication problem Flowers Multiclass Datasets;
   • Data preparation;
   • Dropout test;
   • Fancier Model;
   • Case of Study;
   • Model Training and Metrics:
     • Accuracy
     • Filters
     • Confusion Matrix
     • Hooks
     • Learning Rate Finder

3. Third step - Multiclass_pt2_Lucas_Aires.ipynb

   • Studing the learning rate
   • Changing the value to the LRFinder