Date and time feature extraction in ML Python - Model Pipeline Trace

Model Pipeline - Date and time feature extraction

This pipeline shows how raw date and time data is changed into useful features for a machine learning model. It helps the model understand patterns related to time, like days of the week or hours of the day.

Data Flow - 4 Stages

1Raw data input

1000 rows x 1 column→Dataset with a single column of timestamps (date and time strings)→1000 rows x 1 column

['2024-06-01 14:23:45', '2024-06-02 09:15:00', '2024-06-03 20:05:30']

↓

2Convert to datetime object

1000 rows x 1 column→Parse strings into datetime objects for easier extraction→1000 rows x 1 column

[datetime(2024,6,1,14,23,45), datetime(2024,6,2,9,15,0), datetime(2024,6,3,20,5,30)]

↓

3Feature extraction

1000 rows x 1 column→Extract year, month, day, weekday, hour, minute, second as separate columns→1000 rows x 7 columns

[[2024, 6, 1, 5, 14, 23, 45], [2024, 6, 2, 6, 9, 15, 0], [2024, 6, 3, 0, 20, 5, 30]]

↓

4Model input preparation

1000 rows x 7 columns→Use extracted features as input to train a model→1000 rows x 7 columns

Numerical features ready for model training

Training Trace - Epoch by Epoch

Loss
1.0 |          *
0.8 |        *  
0.6 |      *    
0.4 |    *      
0.2 |  *        
0.0 +-----------
      1 2 3 4 5
      Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	0.85	0.55	Model starts learning with high loss and low accuracy
2	0.65	0.70	Loss decreases and accuracy improves as model learns time patterns
3	0.50	0.78	Model continues to improve with clearer time feature understanding
4	0.40	0.83	Loss lowers steadily, accuracy rises, showing good learning progress
5	0.35	0.86	Model converges with stable loss and high accuracy

Prediction Trace - 9 Layers

Layer 1: Input timestamp

Layer 2: Extract year

Layer 3: Extract month

Layer 4: Extract day

Layer 5: Extract weekday

Layer 6: Extract hour

Layer 7: Extract minute

Layer 8: Extract second

Layer 9: Model prediction

Model Quiz - 3 Questions

What is the shape of the data after extracting date and time features?

A 1000 rows x 1 column

B 700 rows x 7 columns

C 1000 rows x 7 columns

D 1000 rows x 14 columns

Key Insight

Extracting date and time features turns raw timestamps into meaningful numbers that help the model learn patterns related to time, improving prediction accuracy.