InsightAI is a fully offline, rule-based data analytics SaaS platform that turns any raw CSV into a complete business intelligence report. Powered by Pandas, Scikit-Learn, Plotly, and Gradio, it delivers automated data cleaning, statistical analysis, machine learning, forecasting, and business insights — all through a premium, enterprise-style dashboard with zero API keys and zero internet dependency for analysis.
Whether you're exploring a new dataset, prototyping a data science workflow, or showcasing analytics capabilities, InsightAI turns raw rows and columns into instant, decision-ready intelligence.
- 📁 Smart Dataset Profiling — Instant preview, schema overview, dataset health score, and KPI dashboard the moment a CSV is uploaded.
- 🧹 One-Click Data Cleaning — Removes duplicates, fixes missing values, converts date columns, trims whitespace, and drops empty columns automatically.
- 📈 Automatic Visualizations — Bar, line, pie, scatter, histogram, box, violin, correlation heatmap, and pair plots — auto-selected based on your column types.
- 🧠 Rule-Based Insight Engine — Generates human-readable, analyst-style business insights and actionable recommendations without any external AI API.
- 📐 Deep Statistical Analysis — Descriptive statistics, skewness & kurtosis, Shapiro-Wilk normality testing, and IQR/Z-score outlier detection.
- 🤖 Machine Learning Toolkit — K-Means clustering, Linear Regression, Decision Trees, Random Forest feature importance, and PCA — all running locally via Scikit-Learn.
- 🔮 Time Series Forecasting — Auto-detects date columns and projects future trends using Holt's Linear Trend model, complete with growth rate and seasonality detection.
- 📤 One-Click Report Export — Download the cleaned dataset, a polished business PDF report, or a plain-text summary report.
- 🎨 Premium Dashboard UI — A dark, glassmorphism-inspired SaaS interface with a fixed sidebar navigation, gradient accents, and smooth micro-animations.
- 🛡️ Built-In Error Handling — Gracefully handles empty files, unsupported formats, missing columns, and large datasets.
| Tool | Purpose |
|---|---|
| Gradio | Interactive web dashboard |
| Pandas | Data wrangling & cleaning |
| NumPy | Numerical operations |
| Plotly | Interactive visualizations |
| Scikit-Learn | Clustering, regression, trees, PCA |
| SciPy | Statistical testing |
| Statsmodels | Time series forecasting |
| ReportLab | PDF report generation |
| Python 3.9+ | Core application logic |
git clone https://github.com/<your-username>/insightai.git
cd insightaipip install -r requirements.txtpython app.py- Upload any CSV file through the Dashboard page.
- Review the auto-generated dataset health score, KPIs, and schema overview.
- Run the one-click cleaning pipeline from the Settings page.
- Explore auto-generated visualizations and the correlation matrix.
- Generate business insights and recommendations from the rule-based engine.
- Run clustering, regression, decision trees, or PCA on your data.
- Forecast future trends if your dataset contains a date column.
- Export your cleaned dataset, PDF report, or summary report from the Reports page.
The entire pipeline runs locally — no data ever leaves your session.
Tracks:
- Rows & Columns
- Numeric & Categorical Features
- Missing Values & Duplicate Rows
- Memory Usage
- Data Quality Score
- Correlation Heatmap & Pair Plots
- Distribution Charts (Histogram, Box, Violin)
- Category Breakdown (Bar, Pie, Top Categories)
- Forecast Trend Charts with Moving Average
- K-Means Cluster Visualization (PCA-projected)
- Linear Regression Predicted vs. Actual Plot
- Decision Tree Confusion Matrix / Regression Fit
- Random Forest Feature Importance Ranking
- PCA 2D Projection with Explained Variance
- Multi-file dataset comparison mode
- Natural-language query bar for filtering rows
- Auto-generated PowerPoint export
- Support for Excel (
.xlsx) and JSON ingestion - User-saved dashboard templates
- Advanced anomaly detection models
Areef Rasool
BSAI Student | AI, Data Science & Networking Enthusiast
This project is open-source and available under the MIT License.






