Coder Social home page Coder Social logo

olliang / statistical-similarity-measurement Goto Github PK

View Code? Open in Web Editor NEW
9.0 2.0 0.0 3 MB

A methodology designed to validate the statistical similarity of synthetic data generated by GAN models. The metrics contain Auto-encoder, PCA, t-SNE, KL-divergence, Clustering, and Cosine Similarity.

Jupyter Notebook 100.00%
synthetic-data similarity-measures methodology similarity-metric

statistical-similarity-measurement's Introduction

Statistical Similarity Evaluation Metrics for Tabular data

Introduction:

This github is triggered by a research project that I have been working on this spring for a healthcare company. Because of the industry itself, data privacy policy has hindered data from being shared and used efficiently for further analytics and innovation. The business goal of the project is to solve the conflict between data privacy and data sharing. In other words, how can we generate synthetic data that is as similar and useful as original data for the intended use?ย 

Apart from developing powerful synthetic data generating techniques using cutting-edge deep learning algorithms such as GAN (Generative adversarial network), one of important topics in this project is designing similarity evaluation metrics to evaluate whether the synthetic data generated is statistical and structurally similar as the original data.

Measurement of similarity is a widely desired in real business, especially on tabular data. For example, how do we evaluate the similarity between the forecasting sales and the actual sales in order to improve our forecasting model? Can we come up with a score that quantify the level of similarity for management to make better decisions? What if we also need to capture the similarity on the relationships across multiple variables?

If you are facing similar problems or are interested in similarity measurement on tabular data (even though the concept of yours might be slightly different from measuring the similarity of synthetic data), I hope these metrics I tailored for my project can benefit you to some degree.

Main Files:

Statistical Similarity Methodology.md - Main file that documenting the methodology and result

column-wise evaluation.ipynb- Main notebook script that implement the column-wise evaluation

table-wise evaluation.ipynb- Main notebook script that implement the table-wise evaluation

Clustering_LOS.ipynb- Main notebook script that implement clustering metric for 'Length of Stay' use case

Clustering_Mortality.ipynb- Main notebook script that implement clustering metric for 'Mortality' use case

Datasets Overview

datasets

Reference:

https://blog.keras.io/building-autoencoders-in-keras.html

https://towardsdatascience.com/deep-inside-autoencoders-7e41f319999f

Contact:

If there is anything to be corrected or you have any thoughts to share with me on this topic, please feel free to reach out! It's always pleasure to learn more.

Email: [email protected]

statistical-similarity-measurement's People

Contributors

olliang avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.