DataStory AI
AI-Powered Data Analysis Assistant
Built an AI-powered data analysis assistant that turns raw CSV datasets into cleaned data, automated EDA, visual insights, tuned machine learning models, and clear model comparison reports.
System Architecture
Automated data-analysis architecture documented in the DataStory AI repository.
Dataset intake
A CSV file enters a reusable profiling workflow with basic structure and target checks.
Clean + transform
Missing values, types, duplicates, categorical features, and imbalanced classes are handled.
Automated EDA
Statistical summaries and visual diagnostics reveal distributions, relationships, and anomalies.
Search + comparison
Candidate models are tuned with GridSearchCV and compared using task-appropriate evaluation.
Data story workspace
Reports connect cleaning decisions, visual findings, and model results in one analysis narrative.
A traceable path from raw CSV to EDA, tuned models, and readable reports.
Project Overview
DataStory AI is an AI-powered analytics assistant designed to convert raw CSV datasets into a structured data story, from cleaning and exploration to modeling and reporting.
The project automates core data analysis tasks including preprocessing, exploratory data analysis, visual insight generation, model training, hyperparameter tuning, and model comparison.
It uses machine learning workflows such as GridSearchCV and supports class imbalance handling through SMOTE or class weights, making it useful for practical classification datasets.
The assistant focuses on helping users move from raw data to understandable insights faster, combining automation with clear reports that explain what the data shows and how models perform.
Problem
Raw CSV datasets often require repetitive cleaning, EDA, visualization, and model comparison before useful insights can be shared.
Solution
Built an automated analytics assistant that prepares data, generates visual insights, tunes ML models, handles imbalance, and produces comparison-ready reports.