Marketing teams waste 8+ hours/week building reports from spreadsheets. Upload a CSV or Excel file and get instant dashboards, predictive models, marketing attribution, and downloadable reports.
Live Demo -- try it without installing anything.
Coming Soon — Video walkthrough of Insight Engine's key features including auto-profiling, predictive modeling, and marketing attribution.
Watch Demo (Link will be added when video is ready)
- Quick Start (2 min) — Upload a CSV and get instant insights
- Auto-Profiling (2 min) — See column detection, distributions, and correlations in action
- Predictive Modeling (3 min) — Train a model with SHAP explanations in one click
- Marketing Attribution (3 min) — Four attribution models reveal true ROI
- Manual reporting burns time -- Auto-profiler detects column types, distributions, outliers, and correlations in seconds
- No visibility into which channels drive conversions -- Four attribution models show exactly where marketing budget should go
- Predictive modeling requires ML expertise -- Upload labeled data, pick a target column, get a trained model with SHAP explanations
- No way to segment customers -- K-means and DBSCAN clustering with silhouette scoring for automatic customer segmentation
- Forecasting requires specialized tools -- Moving average, exponential smoothing, and ensemble forecasts from any time series column
- Statistical validation is manual -- Automated hypothesis testing selects the right test based on data characteristics
Quantified outcomes from production BI dashboard deployments:
| Metric | Before | After | Improvement |
|---|---|---|---|
| Quarterly report time | 2 weeks | 10 minutes | 99% faster |
| Lead-to-Conversion | 12% | 28% | 133% increase |
| Marketing attribution clarity | Manual guesswork | 4-model analysis | Data-driven decisions |
| Customer churn prediction | Reactive | 85% accuracy | Proactive retention |
| Forecast preparation | 3 days | 30 minutes | 98% reduction |
| Component | Range |
|---|---|
| BI Engine Investment | $5,000 - $10,000 |
| Implementation Timeline | 4 - 8 weeks |
| This Case Study | $7,500 / 6 weeks |
- <2 seconds to profile 100K rows -- Instant insights without waiting for data teams
- 4 attribution models reveal true ROI -- First-touch, last-touch, linear, and time-decay models show which channels actually drive conversions
- SHAP explanations build trust -- Every prediction shows which features matter and why
- Automated statistical testing -- Right test selected automatically based on data characteristics
| Industry | Use Case | Outcome |
|---|---|---|
| E-commerce | Revenue attribution across channels | 35% better budget allocation, 133% conversion lift |
| SaaS | Customer churn prediction and retention | 85% prediction accuracy |
| Marketing | Campaign performance dashboards | 99% faster reporting |
| Finance | Revenue forecasting and anomaly detection | 98% reduction in forecast prep |
| HR | Attrition prediction and risk scoring | Proactive retention strategies |
| Metric | Value |
|---|---|
| Test Suite | 520+ automated tests |
| Auto-Profile Speed | <2s for 100K row CSV |
| Supported Models | 8+ ML algorithms |
| Statistical Tests | 6 hypothesis tests |
| Attribution Models | 4 multi-touch models |
| Explainability | SHAP + feature importance |
| BI Engine Pricing | $5,000 - $10,000 |
| Implementation Timeline | 4 - 8 weeks |
| Proven Conversion Lift | 133% |
- Service 8: Interactive Business Intelligence Dashboards
- Service 9: Automated Reporting Pipelines
- Service 10: Predictive Analytics and Lead Scoring
- Service 16: Marketing Attribution and ROI Analysis
- Google Data Analytics Certificate
- IBM Business Intelligence Analyst Professional Certificate
- Microsoft Data Visualization Professional Certificate
- Microsoft Generative AI for Data Analysis Professional Certificate
- Google Business Intelligence Professional Certificate
- Google Advanced Data Analytics Certificate
flowchart TB
Upload["CSV / Excel Upload"]
Upload --> TypeDetect["Auto-Type Detection"]
TypeDetect -->|numeric| Profiler
TypeDetect -->|categorical| Profiler
TypeDetect -->|datetime| Profiler
TypeDetect -->|text| Profiler
Profiler["Auto-Profiler
statistics, distributions,
correlations, outliers"]
Profiler --> Forecast["Forecasting
ARIMA, Prophet-like,
exponential smoothing"]
Profiler --> Cluster["Clustering
K-Means, DBSCAN,
hierarchical"]
Profiler --> Anomaly["Anomaly Detection
isolation forest,
Z-score, IQR"]
Profiler --> Attrib["Attribution Models
first-touch, last-touch,
linear, time-decay"]
Forecast --> Observatory["Model Observatory
SHAP explainability,
feature importance"]
Cluster --> Observatory
Anomaly --> Observatory
Profiler --> StatTest["Statistical Testing
t-test, chi-square,
ANOVA, Mann-Whitney"]
Profiler --> KPI["KPI Framework
custom metrics,
threshold alerting"]
Profiler --> RegDiag["Regression Diagnostics
residuals, VIF,
heteroscedasticity"]
Profiler --> DQ["Data Quality Scoring
completeness, validity,
consistency checks"]
Observatory --> Dashboard["Streamlit Dashboard
Plotly charts, auto-layout,
PDF/Markdown reports"]
StatTest --> Dashboard
KPI --> Dashboard
RegDiag --> Dashboard
DQ --> Dashboard
Attrib --> Dashboard
| Module | File | Description |
|---|---|---|
| Profiler | profiler.py |
Auto-detect column types, distributions, outliers, and correlations |
| Dashboard Generator | dashboard_generator.py |
Plotly histograms, pie charts, heatmaps, scatter matrices |
| Data Cleaner | cleaner.py |
Dedup (exact + fuzzy), column standardization, smart imputation |
| Predictor | predictor.py |
Auto-detect classification/regression, gradient boosting, SHAP |
| Attribution | attribution.py |
First-touch, last-touch, linear, time-decay marketing attribution |
| Report Generator | report_generator.py |
Markdown reports with findings, metrics, chart placeholders |
| Anomaly Detector | anomaly_detector.py |
Z-score and IQR outlier detection |
| Advanced Anomaly | advanced_anomaly.py |
Isolation forest, LOF, multi-method ensemble detection |
| Clustering | clustering.py |
K-means and DBSCAN with silhouette scoring and cluster comparison |
| Feature Lab | feature_lab.py |
Feature scaling, encoding, polynomial features, interaction terms |
| Forecaster | forecaster.py |
Moving average, exponential smoothing, linear trend, ensemble forecasts |
| Statistical Tests | statistical_tests.py |
t-test, chi-square, ANOVA, Mann-Whitney, Kruskal-Wallis, Shapiro-Wilk |
| KPI Framework | kpi_framework.py |
Custom KPI definitions, threshold alerting, trend tracking |
| Model Observatory | model_observatory.py |
SHAP explanations, feature importance, model comparison |
| Hypertuner | hypertuner.py |
Automated hyperparameter tuning with cross-validation |
| Dimensionality | dimensionality.py |
PCA, t-SNE dimensionality reduction and visualization |
| Regression Diagnostics | regression_diagnostics.py |
Residual analysis, VIF, heteroscedasticity testing |
| Data Quality | data_quality.py |
Completeness, validity, and consistency scoring |
git clone https://github.com/ChunkyTortoise/insight-engine.git
cd insight-engine
pip install -r requirements.txt
make test
make demodocker compose up
# Open http://localhost:8501| Dataset | Rows | Use Case |
|---|---|---|
| E-commerce Transactions | 1,000 | Revenue analysis, category distributions, return rates |
| Marketing Touchpoints | ~800 | Attribution modeling across 6 channels |
| HR Attrition | 500 | Predictive modeling (who will leave?) |
| Layer | Technology |
|---|---|
| UI | Streamlit, Plotly |
| Data | Pandas, NumPy, openpyxl |
| ML | scikit-learn, XGBoost, SHAP |
| Testing | pytest (520+ tests) |
| CI | GitHub Actions (Python 3.11, 3.12) |
| Linting | Ruff |
| Container | Docker, Docker Compose |
insight-engine/
├── app.py # Streamlit application
├── insight_engine/
│ ├── profiler.py # Auto-profiling + column type detection
│ ├── dashboard_generator.py # Chart generation + layout
│ ├── attribution.py # 4 marketing attribution models
│ ├── predictor.py # Auto-ML + SHAP explanations
│ ├── cleaner.py # Dedup, standardize, impute
│ ├── report_generator.py # Markdown/PDF report generation
│ ├── anomaly_detector.py # Z-score + IQR outlier detection
│ ├── advanced_anomaly.py # Isolation forest, LOF, ensemble
│ ├── clustering.py # K-means, DBSCAN, silhouette scores
│ ├── feature_lab.py # Feature scaling, encoding, polynomials
│ ├── forecaster.py # Time series forecasting (4 methods)
│ ├── statistical_tests.py # 6 hypothesis tests
│ ├── kpi_framework.py # KPI definitions and alerting
│ ├── model_observatory.py # SHAP + feature importance
│ ├── hypertuner.py # Hyperparameter tuning
│ ├── dimensionality.py # PCA, t-SNE reduction
│ ├── regression_diagnostics.py # Residual analysis, VIF
│ └── data_quality.py # Quality scoring
├── benchmarks/ # Performance benchmarks
├── demo_data/ # 3 sample datasets
├── docs/adr/ # Architecture Decision Records
├── tests/ # 19 test files, 520+ tests
├── .github/workflows/ci.yml # CI pipeline
├── Dockerfile # Container image
├── docker-compose.yml # Container orchestration
├── Makefile # demo, test, lint, setup
└── requirements.txt
| ADR | Title | Status |
|---|---|---|
| ADR-0001 | Automatic Type Detection | Accepted |
| ADR-0002 | Four Attribution Models | Accepted |
| ADR-0003 | SHAP Explainability | Accepted |
make test # Full suite (520+ tests)
python -m pytest tests/ -v # Verbose output
python -m pytest tests/test_profiler.py # Single moduleSee BENCHMARKS.md for detailed performance data.
python benchmarks/run_benchmarks.py
# Results written to benchmarks/RESULTS.md- EnterpriseHub -- Real estate AI platform with BI dashboards and CRM integration
- docqa-engine -- RAG document Q&A with hybrid retrieval and prompt engineering lab
- ai-orchestrator -- AgentForge: unified async LLM interface (Claude, Gemini, OpenAI, Perplexity)
- scrape-and-serve -- Web scraping, price monitoring, Excel-to-web apps, and SEO tools
- prompt-engineering-lab -- 8 prompt patterns, A/B testing, TF-IDF evaluation
- llm-integration-starter -- Production LLM patterns: completion, streaming, function calling, RAG, hardening
- Portfolio -- Project showcase and services
See CHANGELOG.md for release history.
If Insight Engine has been useful to you, consider sponsoring its continued development:
See SPONSORS.md for sponsorship tiers and benefits.
MIT -- see LICENSE for details.
Need help with data analytics or BI dashboards? I help teams build production data pipelines:
- 📊 Consulting — Data architecture, attribution modeling, dashboard design
- 🚀 Implementation — Predictive models, automated reporting, analytics pipelines
- 📧 Enterprise — Custom integrations, SLAs, dedicated support
See what clients say about working with me: TESTIMONIALS.md
"We used to spend 2 weeks on quarterly reports. Now it's automatic — takes 10 minutes."
— CFO, B2B SaaS Company