This repository contains code for training an exercise classification model using Flux in Julia. The model can classify exercises into different body parts.
The exercise classification model is built using the Flux machine learning library in Julia. It takes a dataset of exercise names and their corresponding body parts as input, and learns to classify exercises into different body parts based on their names.
The code follows the following steps:
-
Data Preprocessing: The dataset is loaded from a CSV file ("megaGymDataset.csv"). Any necessary preprocessing steps, such as handling missing values or encoding categorical variables, are performed.
-
Feature Extraction: The exercise names are one-hot encoded to represent them as numerical features. The target body parts are encoded as integers.
-
Model Training: The data is split into training and testing sets. The model architecture is defined using dense layers. The model is trained using the training data and the cross-entropy loss function.
-
Model Evaluation: The trained model is evaluated on the testing set to calculate the accuracy of the predictions.
-
Model Deployment: The trained model can be used to make predictions on new exercise names. An example exercise name is provided, and the model predicts the corresponding body part.
The code assumes the availability of a CSV dataset file named "megaGymDataset.csv". The dataset should contain exercise names in the "Title" column and corresponding body parts in the "BodyPart" column.