Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis

Code to replicate results in, "Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis."

If you find this work useful, please use the following citation information in plaintext or BibTeX format:

Sisodia, A, Burnap, A, and Kumar, V (2024). Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis. Journal of Marketing Research (under review).

@article{sisodia2024disentangelment,
  title= Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis},
  author={Sisodia, A, Burnap, A, and Kumar, V},
  journal={Journal of Marketing Research (under review)},
  volume={XXX},
  number={XXX},
  pages={XXX},
  year={2024},
  publisher={XXX}
}

Installation

Libraries

This codebase was built using the Pytorch neural net framework. However, it also depends on over 100 other Python packages, as well as GPU libraries including CUDA runtimes and cuDNN kernel support. We can not guarantee this code will run and replicate unless the same environment and library versions are used. To install these packages (including the same library versions), please use the following:

conda env create -f disentanglement_env.yml

Data

To download the dataset, please use the following links. Copy these files to ‘./hyperparameter_selection/data/watches/‘ and ‘./post_model_search/data/watches/‘ directories.

• Watches Dataset

Run

Use python main.py <param-name> <param-value>. For example:

python main.py --sup_signal continuousprice -s 1 --name continuousprice_s1 --btcvae-B 1 --btcvae-M 16

Output

This will create a directory results/<model-name>/ which will contain:

• specs.json: The parameters used to run the program (default and modified with CLI).
• train_losses.csv: All (sub-)losses computed during training on the train and validation dataset.
• test_losses.log: All (sub-)losses computed at the end of training on the test1 and test2 dataset.
• reconstruct_traverse.png: latent traversals of latent dimensions.
• filename_test1.csv: filenames of all watches in the test1 dataset.
• filename_test2.csv: filenames of all watches in the test2 dataset.
• filename_train.csv: filenames of all watches in the train dataset.
• mean_params_test1.csv: mean visual characteristics of all watches in the test1 dataset.
• mean_params_test2.csv: mean visual characteristics of all watches in the test2 dataset.
• mean_params_train.csv: mean visual characteristics of all watches in the train dataset.

Help

usage: main.py ...

General options:
  name Name of the model for storing or loading purposes.
  -s, --seed SEED Random seed.
  -e, --epochs EPOCHS Maximum number of epochs to run for.
  -b, --batch-size BATCH_SIZE Batch size for training.
    --lr LR Learning rate.
  -z, --latent-dim LATENT_DIM Dimension of the latent variable.
  —tv --threshold-val Threshold for Masking.
  --sup_signal Choice of Signal. (Examples: discreteprice, brand, circa, material, movement or a combination in the format <signal1>_<signal2> etc.)

Loss specific hyperparameters:
  --btcvae-A BTCVAE_A			Weight of the MI term (alpha in the paper).
  --btcvae-G BTCVAE_G			Weight of the dim-wise KL term (gamma in the paper).
  --btcvae-B BTCVAE_B			Weight of the TC term (beta in the paper).
  --btcvae-M BTCVAE_M			Weight of the supervised loss term (delta in the paper).

Hyperparameter Selection

Run the same model configuration (i.e. a combination of the supervisory signal and loss-specific hyperparameters) for different random seeds. Select the optimal loss-specific hyperparameters for a particular supervisory signal combination based on the lowest supervised loss on the validation dataset (averaged across different seeds).

UDR Calculation

Run the most optimal loss-specific hyperparameters for each combination of the supervisory signal across different random seeds. Copy the files stored in ‘results//‘ directory with the filename ending in ‘mean_params_test2.csv’ to the ‘calculate_udr’ folder. Switch to an R environment and calculate UDR using the below command. You may need to modify the R script and adjust the model name.

Rscript udr_calculation.R --sup_signal='brand'

Conjoint Analysis and "Ideal Point'' Generative Design

See the directory /conjoint_analysis_and_ideal_point_design for an example ipython notebook for running the hiearachical Bayeisan estimation and replicate results.

Note that generating the ideal point design requires inputting the "ideal point" embedding values into the generative model from the disentanglement portion of this codebase.

Citation

Feel free to use this code for your research. If you find this code useful, please use the following citation information in plaintext or BibTeX format:

Sisodia, A, Burnap, A, and Kumar, V (2024). Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis. Journal of Marketing Research (under review).

@article{sisodia2024disentangelment,
  title= Generative Interpretable Visual Design: Using Disentanglement for Visual Conjoint Analysis},
  author={Sisodia, A, Burnap, A, and Kumar, V},
  journal={Journal of Marketing Research (under review)},
  volume={XXX},
  number={XXX},
  pages={XXX},
  year={2024},
  publisher={XXX}
}

Acknowledgments

Portions of this codebase were built on elements from the following open source projects, please see their respective licenses and cite accordingly:

• disentangling-vae