Simple neural network with scikit-learn in ML Python - Model Pipeline Trace

Model Pipeline - Simple neural network with scikit-learn

This pipeline shows how a simple neural network learns to classify data using scikit-learn. It starts with raw data, prepares it, trains the model, and then makes predictions.

Data Flow - 5 Stages

1Raw Data

150 rows x 4 columns→Load Iris dataset with 4 features per sample→150 rows x 4 columns

[5.1, 3.5, 1.4, 0.2]

↓

2Train/Test Split

150 rows x 4 columns→Split data into 120 training and 30 testing samples→120 rows x 4 columns (train), 30 rows x 4 columns (test)

[5.0, 3.6, 1.4, 0.2] (train), [6.7, 3.1, 4.7, 1.5] (test)

↓

3Feature Scaling

120 rows x 4 columns→Scale features to zero mean and unit variance→120 rows x 4 columns

[-0.9, 1.2, -1.3, -1.4]

↓

4Model Training

120 rows x 4 columns→Train MLPClassifier with 1 hidden layer of 5 neurons→Trained model

Model weights updated after each epoch

↓

5Prediction

30 rows x 4 columns→Model predicts class labels for test samples→30 rows x 1 column (predicted labels)

[0, 1, 2, 1, 0]

Training Trace - Epoch by Epoch

Loss
1.2 |****
0.9 |*** 
0.6 |**  
0.3 |*   
0.0 |    
     1 5 10 15 20 Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	1.2	0.50	Model starts learning, loss is high, accuracy low
5	0.7	0.75	Loss decreases, accuracy improves
10	0.4	0.85	Model is learning well, loss continues to drop
15	0.25	0.90	Loss low, accuracy high, model converging
20	0.15	0.95	Training loss minimal, accuracy near perfect

Prediction Trace - 5 Layers

Layer 1: Input Layer

Layer 2: Feature Scaling

Layer 3: Hidden Layer (ReLU activation)

Layer 4: Output Layer (Softmax)

Layer 5: Prediction

Model Quiz - 3 Questions

What happens to the data shape after feature scaling?

A Number of columns reduces by half

B Number of rows doubles

C Shape stays the same but values change

D Data becomes one-dimensional

Key Insight

This visualization shows how a simple neural network learns by adjusting weights to reduce loss and improve accuracy. Feature scaling helps the model learn faster. The softmax layer turns outputs into probabilities for easy class prediction.