SMOTE & SMOTE-NC Calculator

Machine Learning Lectures SS2026 HKA/EIT

Step-by-step synthetic data generation with full calculations

Minority Samples (T):

Majority Samples:

SMOTE N%:

K Neighbors:

1 Original Imbalanced Dataset

Total Samples: 25

Minority: 5

Majority: 20

Imbalance Ratio: 1:4

Index	X₁ (Feature 1)	X₂ (Feature 2)	Category	Class

Index	X₁ (Feature 1)	X₂ (Feature 2)	Category	Class

Index	X₁ (Feature 1)	X₂ (Feature 2)	Category	Class

📊 Data Visualization: Before Oversampling

Status: Imbalanced

Show Categories (Feature 3)

Feature Space Distribution BEFORE

Minority Class

Majority Class

Minority

5

Majority

20

Ratio

1:4

Class Distribution

⚠ Problem: The minority class is significantly underrepresented, which can lead to biased model predictions favoring the majority class.

4 Synthetic Sample Generation

T (Minority): 5

N%: 200%

To Generate: 5

Generated: 0

SMOTE Algorithm: SMOTE(T, N, k)

Input:
• T = 5 (number of minority class samples)
• N = 200% (amount of SMOTE)
• k = 3 (number of nearest neighbors)
⚠ N < 100%: Applying lines 2-6 of algorithm
Line 3: Randomize the T minority class samples
Line 4: T = (N/100) × T = 1 samples selected
Line 5: N = 100%
(Original T was 5, now only 1 will be SMOTEd)
Output: 10 synthetic samples

                            Line 7: N = (int)(N/100) = 2 → Each selected sample generates 2 synthetic sample(s)
                        

Populate(N, i, nnarray) Function

Synthetic[newindex][attr] = Sample[i][attr] + gap × dif

where: dif = Sample[nnarray[nn]][attr] − Sample[i][attr], gap = random(0, 1)

Click "Run SMOTE" to start generating synthetic samples

📊 Data Visualization: After Oversampling

Status: More Balanced

Feature Space Distribution AFTER

Original Minority

Synthetic Samples

Majority Class

Original

5

Synthetic

5

New Ratio

1:2

Class Distribution Comparison

✓ Result: SMOTE has generated synthetic samples to balance the dataset. The minority class is now better represented, improving model training.

Before vs After Comparison

Improvement Metrics

Minority Increase +100%

Imbalance Reduction 50%

Ratio Before 1:4

Ratio After 1:2