Overview

Personal finance tools are either too heavy or too manual. I built a lightweight tracker that lets users add income and expenses fast, tag categories, and immediately see where money goes. The app shows month-to-date burn-rate, budget progress, and a simple machine-learning forecast for end-of-month spend. Everything runs on a clean Flask backend with a minimal HTML/CSS/JS frontend and Chart.js visuals.

Architecture

• Backend: Flask blueprints (/transactions, /budgets, /analytics). Service layer handles CRUD, summaries, and validation; responses as HTML or JSON for AJAX.
• Data: SQLite with tables: users, transactions (amount, type, category, date, note), budgets (category, month, limit), tags (optional). Indexed on date & category for fast reports.
• Logic: Budget utilization = spent / limit; burn-rate = month-to-date spend / elapsed days; forecast = min( budget_limit, regression( recent_daily_spend ) × days_in_month ).
• Frontend: HTML templates (Jinja) with partials; JS uses fetch to submit forms and refresh charts without page reloads; Chart.js renders line (cashflow), bar (category totals), and doughnut (allocation).
• Security/Validation: CSRF tokens, server-side checks, amount/date sanitization. (Auth optional for personal demo.)
• Deployment: Gunicorn on Render (or Railway), SQLite persisted volume; environment variables for secrets.

Highlights

What I Learned

• Designing APIs that return both HTML (for SSR) and JSON (for AJAX) simplifies progressive enhancement.
• How to structure small Flask apps for growth using blueprints, a service layer, and dependency-free validators.
• Turning simple statistics into useful features (burn-rate, linear regression) adds real value without heavy ML.

Next Steps

• CSV import/export and recurring transactions (rules engine for auto-categorization).
• User auth + roles; multi-budget support; Stripe webhook for automatic income logging.
• Forecast alternatives (exponential smoothing), savings goals, and monthly PDF summary.