ASCVIT

KO | EN

gitlite — search

#python #hacktoberfest #typescript #golang #machine-learning #rust #docker #macos #java #kubernetes #vue #security #mcp #openai

(★ 12)

Repository for the article in the online magazine Towards Data Science.

ASCVIT 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

![](https://raw.githubusercontent.com/stefanpietrusky/ASCVIT/main/images/ASCVIT_Logo.png)

# ASCVIT V3 APP
## Automatic Statistical Calculation, Visualization, and Interpretation Tool

This repository, developed by Stefan Pietrusky, is based on the article published at Towards Data Science [[1]](https://medium.com/towards-data-science/ascvit-v1-automatic-statistical-calculation-visualization-and-interpretation-tool-aa910001a3a7). In this article, I discuss the development of a first version (V1) of a local app that can be used to automatically apply various statistical methods to any datasets. The enhanced version of ASCVIT (V1.5) now includes automatic interpretation of all generated visualisations [appv1.5.py]. The latest version of ASCVIT is now available [appv2.py]. Streamlit is no longer used in this version. The app is completely based on customizable HTML, CSS and JS. An LLM defined via the code is no longer used, but the models installed locally via Ollama can be selected as required. The navigation has been adapted and all data visualizations are now interactive through Plotly.

ASCVIT can be used in social work, specifically in associations and organizations that collect a wide range of data and can use the tool to evaluate it in order to better understand financial needs [[2]](https://www.pedocs.de/frontdoor.php?source_opus=33234). This is an open source project for educational and research purposes.

ASCVIT V3 expands the tool with greater statistical coverage, more automation, and additional guidance on what is possible. The data overview no longer provides just a preview/description, but also automatic recommendations as to which variables and procedures are particularly useful based on missing rate, variance, number of levels/balance, etc. (including detection of date-like columns and binary numerical candidates). This lowers the barrier to entry because “good” analyses are suggested more quickly. Additional procedures have been implemented in the hypothesis testing module. In addition to t-test/ANOVA/Chi², Mann-Whitney U, Welch-ANOVA, Kruskal-Wallis, ANOVA with repeated measurements, and Fisher Exact (2×2) have been added. ASCVIT V3 thus also better covers robust or non-parametric and special designs.

![ASCVIT V1.5 Automatic interpretation by LLM (Image by author)](https://raw.githubusercontent.com/stefanpietrusky/ASCVIT/main/images/ASCVITV2_LLM_Function.png)

> **⚠️ Work in Progress:** This prototyp is currently under active development. While I make it available for research purposes, please be aware that there will be some changes to the functional structure. I recognize that some current technical design decisions may not be optimal and are subject to revision. Researchers using this prototyp should expect potential updates and changes. I recommend checking back regularly for updates and versioning information.

## Overview of the statistical procedures 
The following statistical procedures are supported by the third Version:
![ASCVIT V1.5 Overview of analysis methods (Image by author)](https://raw.githubusercontent.com/stefanpietrusky/ASCVIT/main/images/ASCVITV3_Overview.png)

## Structure and function of ASCVIT [V1.5]
The code to run the app is already in the repository as well as a script (clean.py) to remove missing values from data records. Below is a short GIF showing the structure and function of the app.
![ASCVIT V1.5 (Gif by author](https://raw.githubusercontent.com/stefanpietrusky/ASCVIT/main/images/ASCVITV1.5.gif)

## The latest version of ASCVIT [V3]
The new version of ASCVIT looks as follows.
![ASCVIT V2 (Gif by author](https://raw.githubusercontent.com/stefanpietrusky/ASCVIT/main/images/ASCVITV3.gif)

## Installing and running the application 
1. Clone this repository on your local computer: 
```bash 
git clone https://github.com/stefanpietrusky/ascvitv2.git
```
2. Install the required dependencies depending on the version you want to use:
```bash 
pip install -r requirements.txt
```
3. Install Ollama and load the model [Llama3.1](https://ollama.com/library/llama3.1) (8B) or another.
4. Remove missing data, if available, with the clean.py script
```bash 
python clean.py
```
5. If you use V1.5 start the Streamlit app:
```bash 
streamlit run appv1.5.py
```
6. If you use V2 start app with:
```bash 
python appv2.py
```
7. Use the file [data.csv](https://github.com/stefanpietrusky/ASCVIT/blob/main/data.csv) to test the application.

## References
[1] Pietrusky, S. (2024). ASCVIT V1: Automatic Statistical Calculation, Visualization, and Interpretation Tool. Towards Data Science

[2] Pietrusky, S. (2025). Datafizierung und KI in der Sozialen Arbeit: Automatisierte Auswertung, Visualisierung und Interpretation von Daten zur Optimierung von Prozessen. peDOCs.

ASCVIT

ASCVIT 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

ASCVIT

ASCVIT 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README