407 lines
15 KiB
PHP
407 lines
15 KiB
PHP
<?php
|
||
|
||
namespace App\Http\Controllers;
|
||
|
||
use App\Models\Student;
|
||
use App\Models\PolijeMajor;
|
||
use App\Models\Recommendation;
|
||
use Illuminate\Http\Request;
|
||
use Illuminate\Support\Facades\Auth;
|
||
|
||
class RekomendasiController extends Controller
|
||
{
|
||
public function index()
|
||
{
|
||
$user = Auth::user();
|
||
$student = null;
|
||
if ($user) {
|
||
$student = (object) [
|
||
'user_id' => $user->id,
|
||
'nis' => $user->nis ?? null,
|
||
'kelompok_asal' => $user->kelompok_asal ?? null,
|
||
'foto' => $user->foto ?? null,
|
||
];
|
||
}
|
||
|
||
return view('rekomendasi.input', compact('student'));
|
||
}
|
||
|
||
/**
|
||
* ============================================================
|
||
* ALGORITMA NAIVE BAYES UNTUK REKOMENDASI JURUSAN
|
||
* Sesuai flowchart:
|
||
* 1. Input Data
|
||
* 2. Preprocessing Data
|
||
* 3. Tentukan Hipotesis (H)
|
||
* 4. Hitung Probabilitas Awal (Prior) P(H)
|
||
* 5. Hitung Likelihood P(X|H) per fitur
|
||
* 6. Hitung Probabilitas Gabungan (Rumus Naive Bayes)
|
||
* P(H|X) ∝ P(H) × P(X1|H) × P(X2|H) × ... × P(Xn|H)
|
||
* 7. Klasifikasi (Hasil Rekomendasi)
|
||
* ============================================================
|
||
*/
|
||
public function proses(Request $request)
|
||
{
|
||
$epsilon = 1e-9;
|
||
|
||
// ================================================================
|
||
// LANGKAH 1: INPUT DATA
|
||
// ================================================================
|
||
$scores = $request->only(['mtk', 'fisika', 'kimia', 'biologi', 'ekonomi', 'geografi', 'sosiologi', 'sejarah']);
|
||
$minatRaw = strtolower(trim($request->minat ?? ''));
|
||
$prefStudi = $request->pref_studi ?? 'Sains & Teknologi';
|
||
$citaRaw = strtolower(trim($request->cita_cita ?? ''));
|
||
$prestasiRaw = strtolower(trim($request->prestasi ?? ''));
|
||
|
||
// ================================================================
|
||
// LANGKAH 2: PREPROCESSING DATA
|
||
// ================================================================
|
||
|
||
// 2a. Hitung rata-rata nilai
|
||
$validScores = array_filter($scores, fn($v) => $v !== null && $v !== '');
|
||
$average = count($validScores) > 0 ? array_sum($validScores) / count($validScores) : 0;
|
||
|
||
// 2b. Kategorisasi nilai
|
||
if ($average >= 85) {
|
||
$katNilai = 'Tinggi';
|
||
} elseif ($average >= 70) {
|
||
$katNilai = 'Sedang';
|
||
} else {
|
||
$katNilai = 'Rendah';
|
||
}
|
||
|
||
// 2c. Skor prestasi
|
||
$prestasiScore = $this->hitungSkorPrestasi($prestasiRaw);
|
||
|
||
// ================================================================
|
||
// LANGKAH 3: TENTUKAN HIPOTESIS (H)
|
||
// H = {Jurusan1, Jurusan2, ..., JurusanN} dari database
|
||
// ================================================================
|
||
$jurusanList = PolijeMajor::all();
|
||
|
||
if ($jurusanList->isEmpty()) {
|
||
return back()->with('error', 'Data jurusan belum tersedia di database.');
|
||
}
|
||
|
||
$jumlahJurusan = $jurusanList->count();
|
||
|
||
// ================================================================
|
||
// LANGKAH 4: HITUNG PROBABILITAS AWAL (PRIOR) P(H)
|
||
// Prior uniform: P(H) = 1 / jumlah_jurusan
|
||
// ================================================================
|
||
$prior = 1 / $jumlahJurusan;
|
||
|
||
// ================================================================
|
||
// LANGKAH 5 & 6: HITUNG LIKELIHOOD DAN PROBABILITAS GABUNGAN
|
||
// Rumus Naive Bayes:
|
||
// P(H|X) ∝ P(H) × P(X1|H) × P(X2|H) × P(X3|H) × P(X4|H) × P(X5|H)
|
||
//
|
||
// Fitur (Xi):
|
||
// X1 = Nilai Akademik → P(nilai|H)
|
||
// X2 = Minat → P(minat|H)
|
||
// X3 = Preferensi Studi → P(pref|H)
|
||
// X4 = Cita-cita → P(cita|H)
|
||
// X5 = Prestasi → P(prestasi|H)
|
||
//
|
||
// Weighted Naive Bayes (log-space):
|
||
// log P(H|X) = log P(H) + Σ wi × log P(Xi|H)
|
||
//
|
||
// Bobot (wi):
|
||
// w1 = 0.40 (Nilai), w2 = 0.35 (Minat), w3 = 0.15 (Pref),
|
||
// w4 = 0.05 (Cita-cita), w5 = 0.05 (Prestasi)
|
||
// ================================================================
|
||
$weights = [
|
||
'nilai' => 0.40,
|
||
'minat' => 0.35,
|
||
'pref' => 0.15,
|
||
'cita' => 0.05,
|
||
'prestasi' => 0.05,
|
||
];
|
||
|
||
$logPosteriors = [];
|
||
$detailPerJurusan = [];
|
||
|
||
foreach ($jurusanList as $jurusan) {
|
||
// --- Log Prior ---
|
||
$logPrior = log(max($prior, $epsilon));
|
||
|
||
// --- X1: Likelihood Nilai Akademik P(nilai|H) ---
|
||
$pNilai = $this->hitungLikelihoodNilai($scores, $jurusan->bobot_mapel);
|
||
|
||
// --- X2: Likelihood Minat P(minat|H) ---
|
||
$pMinat = $this->hitungLikelihoodMinat($minatRaw, $jurusan->keywords);
|
||
|
||
// --- X3: Likelihood Preferensi Studi P(pref|H) ---
|
||
$pPref = $this->hitungLikelihoodPref($prefStudi, $jurusan->preferensi_studi);
|
||
|
||
// --- X4: Likelihood Cita-cita P(cita|H) ---
|
||
$pCita = $this->hitungLikelihoodCitaCita($citaRaw, $jurusan->keywords);
|
||
|
||
// --- X5: Likelihood Prestasi P(prestasi|H) ---
|
||
$pPrestasi = $this->hitungLikelihoodPrestasi($prestasiScore);
|
||
|
||
// --- Probabilitas Gabungan (Weighted Naive Bayes) ---
|
||
// log P(H|X) = log P(H) + w1·log P(X1|H) + w2·log P(X2|H) + ... + w5·log P(X5|H)
|
||
$logPosterior = $logPrior
|
||
+ $weights['nilai'] * log(max($pNilai, $epsilon))
|
||
+ $weights['minat'] * log(max($pMinat, $epsilon))
|
||
+ $weights['pref'] * log(max($pPref, $epsilon))
|
||
+ $weights['cita'] * log(max($pCita, $epsilon))
|
||
+ $weights['prestasi'] * log(max($pPrestasi, $epsilon));
|
||
|
||
$logPosteriors[$jurusan->nama_jurusan] = $logPosterior;
|
||
|
||
// Simpan detail per kriteria untuk tampilan
|
||
$detailPerJurusan[$jurusan->nama_jurusan] = [
|
||
'nilai' => round($pNilai, 4),
|
||
'minat' => round($pMinat, 4),
|
||
'pref' => round($pPref, 4),
|
||
'cita' => round($pCita, 4),
|
||
'prestasi' => round($pPrestasi, 4),
|
||
];
|
||
}
|
||
|
||
// ================================================================
|
||
// LANGKAH 7: KLASIFIKASI (HASIL REKOMENDASI)
|
||
// Konversi log-posterior ke probabilitas menggunakan softmax
|
||
// P(Hk|X) = exp(log Pk) / Σ exp(log Pi)
|
||
// ================================================================
|
||
$maxLog = max($logPosteriors);
|
||
$expVals = [];
|
||
$sumExp = 0.0;
|
||
|
||
foreach ($logPosteriors as $namaJurusan => $lv) {
|
||
$expVals[$namaJurusan] = exp($lv - $maxLog);
|
||
$sumExp += $expVals[$namaJurusan];
|
||
}
|
||
|
||
$hasilAkhir = [];
|
||
foreach ($expVals as $namaJurusan => $val) {
|
||
$prob = $val / max($sumExp, $epsilon);
|
||
$hasilAkhir[] = [
|
||
'jurusan' => $namaJurusan,
|
||
'skor' => round($prob, 4),
|
||
'detail' => $detailPerJurusan[$namaJurusan],
|
||
'kecocokan_nilai' => $katNilai,
|
||
'kecocokan_minat' => $minatRaw,
|
||
'kecocokan_pref' => $prefStudi,
|
||
];
|
||
}
|
||
|
||
// Urutkan berdasarkan skor tertinggi
|
||
usort($hasilAkhir, fn($a, $b) => $b['skor'] <=> $a['skor']);
|
||
|
||
// Ambil data jurusan teratas untuk detail view
|
||
$topJurusan = !empty($hasilAkhir) ? PolijeMajor::where('nama_jurusan', $hasilAkhir[0]['jurusan'] ?? '')->first() : null;
|
||
|
||
// Simpan ke database
|
||
$user = Auth::user();
|
||
$savedRec = null;
|
||
if ($user) {
|
||
$savedRec = Recommendation::create([
|
||
'user_id' => $user->id,
|
||
'mtk' => $request->mtk ?? null,
|
||
'fisika' => $request->fisika ?? null,
|
||
'kimia' => $request->kimia ?? null,
|
||
'biologi' => $request->biologi ?? null,
|
||
'ekonomi' => $request->ekonomi ?? null,
|
||
'geografi' => $request->geografi ?? null,
|
||
'sosiologi' => $request->sosiologi ?? null,
|
||
'sejarah' => $request->sejarah ?? null,
|
||
'minat' => $request->minat ?? null,
|
||
'preferensi_studi' => $request->pref_studi ?? null,
|
||
'cita_cita' => $request->cita_cita ?? null,
|
||
'prestasi' => $request->prestasi ?? null,
|
||
'hasil_rekomendasi' => $hasilAkhir,
|
||
]);
|
||
}
|
||
|
||
// Simpan recommendation_id ke session agar bisa dipakai link chatbot
|
||
$recId = $savedRec ? $savedRec->id : null;
|
||
session(['last_recommendation_id' => $recId]);
|
||
|
||
// Simpan ke session untuk chatbot
|
||
if (count($hasilAkhir) > 0) {
|
||
$topResult = $hasilAkhir[0];
|
||
// Ambil top 3 untuk konteks chatbot
|
||
$top3 = array_slice($hasilAkhir, 0, 3);
|
||
session([
|
||
'recomendation_data' => [
|
||
'jurusan' => $topResult['jurusan'],
|
||
'skor' => $topResult['skor'], // Sudah 0-1, jangan ×100
|
||
'detail' => $topResult['detail'] ?? [],
|
||
'nilai' => $katNilai,
|
||
'rata_rata' => round($average, 1),
|
||
'minat' => $minatRaw,
|
||
'pref_studi' => $prefStudi,
|
||
'cita_cita' => $citaRaw,
|
||
'prestasi' => $prestasiRaw,
|
||
'top3' => array_map(fn($r) => [
|
||
'jurusan' => $r['jurusan'],
|
||
'skor' => $r['skor'],
|
||
], $top3),
|
||
]
|
||
]);
|
||
}
|
||
|
||
return view('rekomendasi.hasil', compact(
|
||
'hasilAkhir', 'katNilai', 'average', 'prefStudi', 'prestasiScore', 'topJurusan'
|
||
));
|
||
}
|
||
|
||
// ==================================================================
|
||
// FUNGSI LIKELIHOOD — P(Xi | H)
|
||
// ==================================================================
|
||
|
||
/**
|
||
* P(nilai | H) — Likelihood nilai akademik terhadap jurusan
|
||
* Menggunakan bobot_mapel dari database untuk menghitung
|
||
* weighted average yang dinormalisasi ke range probabilitas.
|
||
*/
|
||
private function hitungLikelihoodNilai(array $scores, ?array $bobotMapel): float
|
||
{
|
||
// Jika tidak ada bobot, gunakan rata-rata biasa
|
||
if (empty($bobotMapel)) {
|
||
$valid = array_filter($scores, fn($v) => $v !== null && $v !== '');
|
||
if (empty($valid)) return 0.3;
|
||
$avg = array_sum($valid) / count($valid);
|
||
return $this->normalisasiProbabilitas($avg / 100, 0.10, 0.95);
|
||
}
|
||
|
||
$weightedSum = 0;
|
||
$totalWeight = 0;
|
||
|
||
foreach ($bobotMapel as $subject => $weight) {
|
||
$nilai = floatval($scores[$subject] ?? 0);
|
||
if ($nilai > 0 && $weight > 0) {
|
||
$weightedSum += $weight * ($nilai / 100);
|
||
$totalWeight += $weight;
|
||
}
|
||
}
|
||
|
||
if ($totalWeight == 0) return 0.3;
|
||
|
||
$weightedAvg = $weightedSum / $totalWeight;
|
||
return $this->normalisasiProbabilitas($weightedAvg, 0.10, 0.95);
|
||
}
|
||
|
||
/**
|
||
* P(minat | H) — Likelihood minat terhadap jurusan
|
||
* Menggunakan keyword matching terhadap keywords jurusan dari database.
|
||
*/
|
||
private function hitungLikelihoodMinat(string $minatRaw, ?array $keywords): float
|
||
{
|
||
if (empty($keywords) || empty($minatRaw)) {
|
||
return 0.20; // probabilitas dasar jika tidak ada data
|
||
}
|
||
|
||
$matchCount = 0;
|
||
foreach ($keywords as $keyword) {
|
||
if (stripos($minatRaw, strtolower($keyword)) !== false) {
|
||
$matchCount++;
|
||
}
|
||
}
|
||
|
||
// Rasio kecocokan keyword
|
||
$matchRatio = $matchCount / count($keywords);
|
||
|
||
// Konversi ke range probabilitas: 0 match → 0.10, full match → 0.95
|
||
return $this->normalisasiProbabilitas($matchRatio, 0.10, 0.95);
|
||
}
|
||
|
||
/**
|
||
* P(pref | H) — Likelihood preferensi studi terhadap jurusan
|
||
* Membandingkan preferensi siswa dengan preferensi_studi jurusan dari database.
|
||
*/
|
||
private function hitungLikelihoodPref(string $prefStudi, ?array $jurusanPref): float
|
||
{
|
||
if (empty($jurusanPref)) {
|
||
return 0.40; // probabilitas netral
|
||
}
|
||
|
||
// Cek apakah preferensi siswa ada di list preferensi jurusan
|
||
if (in_array($prefStudi, $jurusanPref)) {
|
||
return 0.85; // cocok
|
||
}
|
||
|
||
return 0.15; // tidak cocok
|
||
}
|
||
|
||
/**
|
||
* P(cita_cita | H) — Likelihood cita-cita terhadap jurusan
|
||
* Menggunakan keyword matching dari cita-cita siswa terhadap keywords jurusan.
|
||
*/
|
||
private function hitungLikelihoodCitaCita(string $citaRaw, ?array $keywords): float
|
||
{
|
||
if (empty($keywords) || empty($citaRaw)) {
|
||
return 0.25; // probabilitas dasar
|
||
}
|
||
|
||
$matchCount = 0;
|
||
foreach ($keywords as $keyword) {
|
||
if (stripos($citaRaw, strtolower($keyword)) !== false) {
|
||
$matchCount++;
|
||
}
|
||
}
|
||
|
||
$matchRatio = $matchCount / count($keywords);
|
||
return $this->normalisasiProbabilitas($matchRatio, 0.10, 0.90);
|
||
}
|
||
|
||
/**
|
||
* P(prestasi | H) — Likelihood prestasi
|
||
* Prestasi bersifat umum (tidak spesifik per jurusan), sehingga
|
||
* memberikan boost yang sama untuk semua jurusan.
|
||
*/
|
||
private function hitungLikelihoodPrestasi(float $prestasiScore): float
|
||
{
|
||
// Konversi skor prestasi (0-1) ke range probabilitas
|
||
return $this->normalisasiProbabilitas($prestasiScore, 0.20, 0.90);
|
||
}
|
||
|
||
// ==================================================================
|
||
// FUNGSI HELPER
|
||
// ==================================================================
|
||
|
||
/**
|
||
* Normalisasi nilai (0-1) ke range probabilitas [min, max]
|
||
* Agar tidak ada likelihood 0 atau 1 (menghindari dominasi)
|
||
*/
|
||
private function normalisasiProbabilitas(float $value, float $min = 0.10, float $max = 0.95): float
|
||
{
|
||
return $min + ($value * ($max - $min));
|
||
}
|
||
|
||
/**
|
||
* Hitung skor prestasi berdasarkan keyword
|
||
*/
|
||
private function hitungSkorPrestasi(string $prestasiRaw): float
|
||
{
|
||
if (empty($prestasiRaw)) {
|
||
return 0.0;
|
||
}
|
||
|
||
if (preg_match('/(juara|menang|champion|first|gold|emas|terbaik)/', $prestasiRaw)) {
|
||
return 0.90;
|
||
} elseif (preg_match('/(finalis|semifinal|peringkat|ranking|podium|medali|silver|perak)/', $prestasiRaw)) {
|
||
return 0.75;
|
||
} elseif (preg_match('/(sertifikat|training|kursus|workshop|peserta|mengikuti)/', $prestasiRaw)) {
|
||
return 0.60;
|
||
}
|
||
|
||
return 0.30;
|
||
}
|
||
|
||
/**
|
||
* Tampilkan history rekomendasi
|
||
*/
|
||
public function historyRekomendasi()
|
||
{
|
||
$user = Auth::user();
|
||
$recommendations = Recommendation::where('user_id', $user->id)
|
||
->orderBy('created_at', 'desc')
|
||
->get();
|
||
|
||
return view('history.rekomendasi', compact('recommendations'));
|
||
}
|
||
} |