The instructions for setting up, running, and testing the different algorithms are separated by dataset.

Dataset 1: Spotify Song Popularity Classifier

TASK: Given features about each song predict the number of streams within a certain range.

Dataset 2: Car Make Classifier

TASK: Given images of cars classify them into the make (manufacturer)

Requirements

• Python 3.6+
• pandas
• torch
• torchvision
• Pillow
• matplotlib
• scikit-learn

Setup and Usage

1. Prepare Data:

   • This is only required when the data is being downloaded for the first time the box folder should contain this data already prepared.
   • Ensure your data is in the following structure:
     • anno_train.csv and train/ for training data.
     • anno_test.csv and test/ for test data.
   • Merge and label the data by running:

     python data_format.py

   • This will create merged_dataset.csv and a folder named train.

2. Verify Data:

   • Run the viewer script from the terminal to verify the merged data:

     python viewr.py

3. Train the Model:

   • Train the model and generate plots by running:

     python train.py

   • Select the tests to be run in the main() function of the train.py script.
   • Te corresponding plots will be generated once training is completed.