Confusion matrix visualization in TensorFlow - Model Pipeline Trace

Model Pipeline - Confusion matrix visualization

This pipeline trains a simple model to classify images into two categories. After training, it creates a confusion matrix to visualize how well the model predicts each class, showing correct and incorrect predictions.

Data Flow - 6 Stages

1Load dataset

None→Load 1000 images with labels (2 classes)→1000 rows x 28x28 pixels x 1 channel

Image of handwritten digit '0' with label 0

↓

2Preprocessing

1000 rows x 28x28 pixels x 1 channel→Normalize pixel values to range 0-1→1000 rows x 28x28 pixels x 1 channel

Pixel value 255 becomes 1.0

↓

3Train/test split

1000 rows x 28x28 pixels x 1 channel→Split data into 800 training and 200 testing samples→Train: 800 rows x 28x28 pixels x 1 channel, Test: 200 rows x 28x28 pixels x 1 channel

Training image with label 1

↓

4Model training

800 rows x 28x28 pixels x 1 channel→Train simple neural network for 5 epochs→Trained model

Model learns to classify images

↓

5Prediction on test set

200 rows x 28x28 pixels x 1 channel→Model predicts class probabilities→200 rows x 2 classes

[0.8, 0.2] predicted probabilities for one image

↓

6Confusion matrix calculation

200 true labels, 200 predicted labels→Compare true and predicted labels to count correct/incorrect predictions→2 x 2 matrix

[[90, 10], [15, 85]]

Training Trace - Epoch by Epoch

Loss
0.7 |****
0.6 |*** 
0.5 |**  
0.4 |**  
0.3 |*   
0.2 |*   
     1 2 3 4 5 Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	0.65	0.60	Model starts learning basic patterns
2	0.45	0.75	Accuracy improves as model adjusts weights
3	0.35	0.82	Model captures more features
4	0.28	0.87	Loss decreases steadily
5	0.22	0.90	Model converges with good accuracy

Prediction Trace - 4 Layers

Layer 1: Input layer

Layer 2: Dense hidden layer with ReLU

Layer 3: Output layer with softmax

Layer 4: Prediction

Model Quiz - 3 Questions

What does the confusion matrix tell us about the model?

A How many predictions were correct or incorrect for each class

B The exact loss value during training

C The number of layers in the model

D The pixel values of input images

Key Insight

The confusion matrix helps us understand not just overall accuracy but also which classes the model confuses. This insight guides improvements by showing specific errors.