Research Code and Results for ICIS 2025 Publication

If you would like to cite my paper:

Liu, Qirui and Garg, Rajiv, "Unveiling NFT Price Dynamics: The Interplay of Traits, Timing, and Social Signaling" (2025). ICIS 2025 Proceedings. 12. https://aisel.aisnet.org/icis2025/fintech/fintech/12

LaTex:

@inproceedings{LiuGarg2025,
  author = {Liu, Qirui and Garg, Rajiv},
  title = {Unveiling {NFT} Price Dynamics: The Interplay of Traits, Timing, and Social Signaling},
  booktitle = {ICIS 2025 Proceedings},
  year = {2025},
  pages = {12},
  url = {https://aisel.aisnet.org/icis2025/fintech/fintech/12},
  note = {[Online]. Available: \url{https://aisel.aisnet.org/icis2025/fintech/fintech/12}}
}

Code Files Structure

NFT_ICIS25/
├── [Analysis](Analysis/)                       # Research analysis notebooks and models
│   ├── [Analysis](Analysis/Analysis/)
│   │   ├── [Final_Models](Analysis/Analysis/Final_Models/)
│   │   │   ├── Model1_Final.ipynb
│   │   │   ├── Model2_0805.ipynb
│   │   │   ├── Model3_Final.ipynb
│   │   │   ├── Panel_for_Model2.csv
│   │   │   ├── df_offer_monthly.csv
│   │   │   ├── df_table1.csv
│   │   │   ├── df_table2.csv
│   │   │   ├── df_table3.csv
│   │   │   ├── df_table4.csv
│   │   │   ├── df_table5.csv
│   │   │   ├── df_table6.csv
│   │   │   └── df_table7.csv
│   │   └── [Results](Analysis/Analysis/Results/)
│   │       ├── 0805Model_Comparison_Table.csv
│   │       ├── best_AdaBoost.joblib
│   │       ├── best_GradientBoosting.joblib
│   │       ├── best_RandomForest.joblib
│   │       ├── best_XGBoost.joblib
│   │       ├── cluster_2_features.csv
│   │       ├── dummies_result.csv
│   │       ├── feature_importance.png
│   │       └── [more model outputs...]
│   │ 
│   └── [EDA](Analysis/EDA/)                    # Exploratory data analysis
│       ├── Distribution_association.ipynb
│       ├── Offer_data_analysis0915.ipynb
│       └── Transfer_clusters.ipynb
│ 
├── [Data](Data/)                               # Raw and processed datasets
│   ├── [Collection](Data/Collection/)          # Data collection scripts
│   │   ├── nft_transaction_data/
│   │   │   ├── EVENT_JSON_format.json
│   │   │   └── nft_event_offer.py
│   │   └── user_address_data/
│   │       ├── Etherscan_User.py
│   │       ├── etherscan_fix.py
│   │       └── rename_address.py
│   └── [Handling](Data/Handling/)              # Data cleaning notebooks
│       └── Table_NFTs.ipynb
│ 
├── LICENSE
│ 
├── requirements.txt
│ 
└── README.md

Detail Guide

Data

Data consists two parts:

Collection (Data/Collection/) contains the code used to collect the data for both NFT Characteristics and Buyer/Seller Characteristics.

• Under nft_transaction_data, the scripts retrieve detailed NFT event information (7 event types; see the OpenSea API documentation for specifics). You must edit nft_event_offer.py to add your own OpenSea API key and configure your preferred file paths before running the code.

• Under user_address_data, the scripts obtain the address information of the Buyer/Seller that appeared in both N and N-1 sales. Table creation is a requirement in order to run any code under this file. Etherscan_User.py is the primary script to obtain the data, and etherscan_fix.py is for validating the data collection and recollecting the missing data due to API Errors. rename_address.py helps you organize the data collected to have its name by correct buyer and seller addresses.

Handling (Data/Handling/) contains the code used for Table Creation (Parsing the very large json files).

• This is very necessary for table creations. Tables are made for various purposes. Please view the detailed Table_NFTs.ipynb for specific information (what each table is for). Note: This section is primarily created for Github Repo showcase, where some portions are different than the original file handling.

In order to do your own data collection, you must go through the steps above. The data (198 GB) is too large for the GitHub Repo to handle.

Analysis

Because the raw data is large, all derived tables are stored in Analysis/Analysis/Final_Models/. This folder also contains the final models for each section of the paper, with results documented in the corresponding .ipynb notebooks. Analysis/EDA/ contains the exploratory data analysis notebooks. The Results folder stores selected saved outputs and artifacts for the final models.

Environment

Install dependencies with:

pip install -r requirements.txt