Moving averages in ML Python - Model Pipeline Trace

Model Pipeline - Moving averages

This pipeline uses moving averages to smooth noisy data. It helps the model see clearer trends by averaging recent points before training.

Data Flow - 4 Stages

1Raw data input

1000 rows x 1 column→Collect raw time series data points→1000 rows x 1 column

[23, 25, 22, 24, 26, 28, 27, 29, 30, 31, ...]

↓

2Calculate moving average

1000 rows x 1 column→Compute 5-point moving average to smooth data→996 rows x 1 column

[24.0, 25.0, 25.4, 26.8, 28.0, 29.0, 29.8, 31.0, 32.0, ...]

↓

3Train/test split

996 rows x 1 column→Split data into 80% training and 20% testing sets→Train: 796 rows x 1 column, Test: 200 rows x 1 column

Train: [24.0, 25.0, ..., 29.0], Test: [29.8, 31.0, ...]

↓

4Model training

Train: 796 rows x 1 column→Train simple regression model on smoothed data→Trained model

Model learns trend from moving average values

Training Trace - Epoch by Epoch

Loss
0.15 |*****
0.10 |****
0.07 |***
0.05 |**
0.04 |*
      +---------
       1 2 3 4 5 Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	0.15	0.60	Model starts learning, loss is high, accuracy low
2	0.10	0.72	Loss decreases, accuracy improves
3	0.07	0.80	Model fits data better, smoother predictions
4	0.05	0.85	Loss continues to drop, accuracy rises
5	0.04	0.88	Training converges, good fit on smoothed data

Prediction Trace - 3 Layers

Layer 1: Input raw data sample

Layer 2: Calculate 5-point moving average

Layer 3: Model prediction

Model Quiz - 3 Questions

What does the moving average step do to the data?

A Randomly changes data values

B Removes all data points

C Smooths the data by averaging recent points

D Splits data into training and testing sets

Key Insight

Using moving averages helps the model see clearer trends by reducing noise. This smoothing improves training stability and prediction accuracy.