Upload 6 files

metrics.txt

@@ -0,0 +1,2 @@
+Accuracy: 0.9758
+Precision: 1.0000

model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f13dc88836745131a01a3882112fc4e1b5cc21c93c8559579e577cc691a3a330
+size 24989568

model_optimized.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b273536c5f5e0fc0f532393afec5ec9a972e14970756c4db828235c222127622
+size 29159702

models.md

@@ -0,0 +1,37 @@
+# Comparison of Training Models
+
+1. **Lite Model**
+- Uses basic ensemble voting with fixed weights [2,1,2]
+- Pre-configured hyperparameters (no optimization)
+- Lemmatization for text preprocessing
+- Uses 3 algorithms: SVC, MultinomialNB, ExtraTreesClassifier
+- Fastest because no parameter tuning/optimization
+
+2. **Legacy Model**
+- Uses simple voting ensemble without weight optimization
+- Porter Stemming for text preprocessing (simpler than lemmatization)
+- Slightly different hyperparameters:
+  - SVC with 'sigmoid' kernel
+  - Fewer trees in ExtraTreesClassifier (50 vs 200)
+- Medium speed due to simpler preprocessing
+
+3. **Monarch Butterfly Optimization (MBO) Model**
+- Uses nature-inspired optimization algorithm
+- Optimizes 7 parameters simultaneously:
+  - SVC parameters (C, gamma)
+  - MultinomialNB alpha
+  - Number of trees
+  - Ensemble weights (w1, w2, w3)
+- Population-based search with:
+  - 20 butterflies
+  - 30 iterations
+  - Cross-validation for each evaluation
+- Slowest because:
+  - Runs multiple training cycles (20 butterflies × 30 iterations)
+  - Each evaluation requires 5-fold cross-validation
+  - Total of ~3000 model evaluations
+
+**Summary**:
+- Lite: Quick, fixed parameters
+- Legacy: Traditional, basic ensemble
+- MBO: Advanced optimization, but computationally intensive

vectorizer.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82b35da250caa705d592c11d661a79f20496f04caa5080c9956fd3de5491a19a
+size 105868

vectorizer_optimized.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82af4b40b45ed7a05dd55d80ba6c063200f797cb7331d7a960c5caa2965316a8
+size 197799