Skip to content

Week 2 Quiz :Sequences, Time Series and Prediction (DeepLearning.AI TensorFlow Developer Professional Certificate) Answers 2025

1. Question 1 — What does MAE stand for?

  • ❌ Mean Average Error

  • ❌ Mean Advanced Error

  • ✅ Mean Absolute Error

  • ❌ Mean Active Error

Explanation:
MAE = average of absolute differences between predictions and actual values.

2. Question 2 — Correct code to split window into features + label

  • ❌ window[n-1], window[1]

  • ✅ dataset = dataset.map(lambda window: (window[:-1], window[-1:]))

  • ❌ window[-1:], window[:-1]

  • ❌ window[n], window[1]

Explanation:
window[:-1] = all except last → features
window[-1:] = last element → label

3. Question 3 — How to inspect learned parameters?

  • ❌ Decompile model

  • ❌ Run model with unit data

  • ❌ Iterate layers blindly

  • ✅ Assign the layer to a variable, add it to the model, inspect after training

Explanation:
If you keep a reference like:

dense = tf.keras.layers.Dense(10)
model.add(dense)
dense.get_weights()

You can inspect weights easily.

4. Question 4 — Change learning rate after each epoch

  • ❌ LearningRateScheduler in model.compile()

  • ❌ Custom callback modifying SGD directly

  • ✅ Use LearningRateScheduler in callbacks of model.fit()

  • ❌ You can’t set it

Explanation:
Use:

callbacks=[tf.keras.callbacks.LearningRateScheduler(fn)]

5. Question 5 — What does drop_remainder=True do?

  • ❌ Adds data

  • ❌ Crops data

  • ❌ Ensures all data is used

  • ✅ Ensures all batches have the same shape (drops incomplete final batch)

Explanation:
If leftover data doesn’t fill a full batch, it gets dropped.

6. Question 6 — What does MSE stand for?

  • ✅ Mean Squared Error

  • ❌ Mean Slight Error

  • ❌ Mean Series Error

  • ❌ Mean Second Error

7. Question 7 — Correct train/validation split

  • ❌ First option

  • ❌ Second option

  • ✅ time_train = time[:split_time]
    x_train = series[:split_time]
    time_valid = time[split_time:]
    x_valid = series[split_time:]

  • ❌ Fourth option

Explanation:
Before split_time → training
After split_time → validation

8. Question 8 — How to set learning rate of SGD optimizer?

  • ❌ RateOfLearning

  • ❌ Can’t set

  • ❌ Rate

  • ✅ learning_rate

Example:

optimizer = tf.keras.optimizers.SGD(learning_rate=1e-4)

9. Question 9 — What is a windowed dataset?

  • ❌ Consistent set of subsets

  • ❌ Time series aligned to a fixed shape

  • ❌ No such thing

  • ✅ A fixed-size subset of a time series

Explanation:
A “window” = a slice of consecutive time steps.

🧾 Summary Table

Q# Correct Answer Key Concept
1 Mean Absolute Error Loss metric
2 (window[:-1], window[-1:]) Window split
3 Assign layer variable & inspect Extracting weights
4 LearningRateScheduler in fit() Dynamic learning rates
5 Drops incomplete final batch drop_remainder
6 Mean Squared Error Popular loss metric
7 Proper slicing for split Train/validation split
8 learning_rate property Optimizer config
9 Fixed-size subset Windowed dataset