ent-embedding 0.0.1

Transforms categorical features into embedded vectors

Package transforms your categorical variables into embedded vectors. You should have tensorflow, pandas, numpy, keras and sklearn installed.

Attributes: model = EntityEmbedding(dataframe, features from the copy of the df, target column, column you want a vector for)

Hyperparameters you can optimize: model.train_fit(activation1='relu', activation2='relu', activation3='relu', loss='mean_squared_error', metrics='mape', dense_size_num=128, dense_size_conc_1=300, dense_size_conc_2=300, alpha=1e-3, epochs=1000, batch_size=512, verbose=1, patience=5)

Inside model.transform(), always provide embedded vector you want to use: model.transform(model.ent_emb)

model.visualize() returns 2 d visualization of your column categories.