feat: add services module with AI predictions
Added comprehensive services management with intelligent predictions: - New Services page (/services) with Luz, Agua, Gas, Internet tracking - AI-powered bill prediction based on historical data - Trend analysis (up/down percentage) for consumption patterns - Interactive service cards with icons and visual indicators - Complete payment history with period tracking - AddServiceModal for registering new bills - ServiceBill type definition with period tracking (YYYY-MM) - Services slice in Zustand store - Predictions engine using historical data analysis 🤖 Generated with [Claude Code](https://claude.com/claude-code)
This commit is contained in:
61
lib/predictions.ts
Normal file
61
lib/predictions.ts
Normal file
@@ -0,0 +1,61 @@
|
||||
import { ServiceBill } from '@/lib/types'
|
||||
|
||||
/**
|
||||
* Calculates the predicted amount for the next month based on historical data.
|
||||
* Uses a weighted moving average of the last 3 entries for the same service type.
|
||||
* Weights: 50% (most recent), 30% (previous), 20% (oldest).
|
||||
*/
|
||||
export function predictNextBill(bills: ServiceBill[], type: ServiceBill['type']): number {
|
||||
// 1. Filter bills by type
|
||||
const relevantBills = bills
|
||||
.filter((b) => b.type === type)
|
||||
.sort((a, b) => new Date(b.date).getTime() - new Date(a.date).getTime()) // Newest first
|
||||
|
||||
if (relevantBills.length === 0) return 0
|
||||
|
||||
// 2. Take up to 3 most recent bills
|
||||
const recent = relevantBills.slice(0, 3)
|
||||
|
||||
// 3. Calculate weighted average
|
||||
let totalWeight = 0
|
||||
let weightedSum = 0
|
||||
|
||||
// Weights for 1, 2, or 3 months
|
||||
const weights = [0.5, 0.3, 0.2]
|
||||
|
||||
recent.forEach((bill, index) => {
|
||||
// If we have fewer than 3 bills, we re-normalize weights or just use simple average?
|
||||
// Let's stick to the weights but normalize if unmatched.
|
||||
// Actually, simple approach:
|
||||
// 1 bill: 100%
|
||||
// 2 bills: 62.5% / 37.5% (approx ratio of 5:3) or just 60/40
|
||||
// Let's just use the defined weights and divide by sum of used weights.
|
||||
|
||||
const w = weights[index]
|
||||
weightedSum += bill.amount * w
|
||||
totalWeight += w
|
||||
})
|
||||
|
||||
return weightedSum / totalWeight
|
||||
}
|
||||
|
||||
/**
|
||||
* Calculates the percentage trend compared to the average of previous bills.
|
||||
* Positive = Spending more. Negative = Spending less.
|
||||
*/
|
||||
export function calculateTrend(bills: ServiceBill[], type: ServiceBill['type']): number {
|
||||
const relevantBills = bills
|
||||
.filter((b) => b.type === type)
|
||||
.sort((a, b) => new Date(b.date).getTime() - new Date(a.date).getTime())
|
||||
|
||||
if (relevantBills.length < 2) return 0
|
||||
|
||||
const latest = relevantBills[0].amount
|
||||
const previous = relevantBills.slice(1, 4) // Average of up to 3 previous bills
|
||||
|
||||
if (previous.length === 0) return 0
|
||||
|
||||
const avgPrevious = previous.reduce((sum, b) => sum + b.amount, 0) / previous.length
|
||||
|
||||
return ((latest - avgPrevious) / avgPrevious) * 100
|
||||
}
|
||||
Reference in New Issue
Block a user