Contenido del Curso
Neural Networks with TensorFlow
Neural Networks with TensorFlow
Challenge
It's time for you to tackle a practical, real-world task involving a substantial amount of data. Your objective is to predict the popularity of tracks on Spotify using its extensive dataset, which includes a variety of features and numerous samples.
Here's a rundown of all the columns in the dataset:
You should start by exploring the Exploratory Data Analysis (EDA) for this dataset:
Your main challenge is to train a model using a portion of this dataset and then make predictions on the test features. Achieving a specific Mean Squared Error (MSE) value is crucial for the success of this task.
You have the flexibility to experiment with various aspects such as the number of epochs, batch size, data preprocessing methods, and learning rate. Apply all the knowledge and techniques you've acquired throughout the course to reach the desired performance level.
Note
To ensure accurate results on the test set, you should initially divide your data into training and validation sets. Aim for the target score on your validation set first. Once achieved, retrain the model on the entire dataset (including both training and validation data) using the same hyperparameters. This approach helps in ensuring the model's robust generalization to new data.
Proceed with the task in Google Colab, and don't forget to upload your predictions once completed.
If you're new to Google Colab, you can check out this example for guidance on how to complete the task using this platform.
¡Gracias por tus comentarios!