PPT by:
Sripooja Mallam
Support Vector Machine (SVM)
Code walkthrough
Neil Gogte
KMIT
KMIT | DL, GenAI | 1
Basic SVM code
PPT by:
Sripooja Mallam
Using numpy and pandas to
implement SVM from scratch
KMIT | DL, GenAI | 2
Basic SVM code
Generate a synthetic cancer dataset
PPT by:
Sripooja Mallam
benign: Samples centered around [2, 2] to
represent benign cases.
malignant: Samples centered around [6, 6]
to represent malignant cases.
KMIT | DL, GenAI | 3
Basic SVM code Implement a basic linear SVM using
gradient descent.
PPT by:
Sripooja Mallam
Update Rules:
● If the condition is satisfied, apply regularization
to the weights.
● If misclassified, update weights and bias to
penalize the misclassification.
KMIT | DL, GenAI | 4
Basic SVM code
PPT by:
Sripooja Mallam
Predict method not invoked in
this code snippet
● Computes the dot product between input features X and learned weights
weights, subtracting the bias.
● Uses the np.sign function to classify:
○ Returns 1 for malignant and -1 for benign.
KMIT | DL, GenAI | 5
● Data Visualization:
Basic SVM code ○ Uses plt.scatter to plot benign and malignant
samples. c=labels ensures different colors for the
two classes. PPT by:
Sripooja Mallam
○ cmap='bwr': A color map where benign points are
blue, and malignant points are red.
● Hyperplane Visualization:
○ y_values = -(weights[0] *
x_values - bias) / weights[1]
calculates the decision
boundary (hyperplane
equation for SVM).
○ Plotted using plt.plot
KMIT | DL, GenAI | 6
Basic SVM code
PPT by:
Sripooja Mallam
KMIT | DL, GenAI | 7
Basic SVM code
PPT by:
Sripooja Mallam
Output is:
The sample [8.35 7.2 ] is classified as: Malignant
KMIT | DL, GenAI | 8
