1. Question 1

Python library used to read CSV files?

• ❌ py_csv

• ❌ py_files

• ❌ CommaSeparatedValues

• ✅ csv

Explanation:
Python’s built-in csv module is used for reading/writing CSV files.

---

2. Question 2

Why is MAE good for accuracy in time series?

• ❌ It punishes larger errors

• ✅ It doesn’t heavily punish larger errors like squared errors do

• ❌ Only counts positive errors

• ❌ Biases toward small errors

Explanation:
MAE treats all errors linearly → good when you don’t want large errors exaggerated (unlike MSE).

---

3. Question 3

Expected input shape for a univariate time series to Conv1D?

• ✅ (window_size, 1)

• ❌ (1, window_size)

• ❌ ()

• ❌ (1,)

Explanation:
Conv1D expects:
(timesteps, features)
Univariate = 1 feature.

---

4. Question 4

Correct way to cast column 2 to float?

• ❌ Convert.toFloat(row[2])

• ❌ float f = row[2].read()

• ✅ float(row[2])

• ❌ You can’t

Explanation:
Casting strings from CSV to float uses:

---

5. Question 5

How to add a 1-D convolution layer?

• ❌ ConvolutionD1

• ❌ 1DConvolution

• ✅ Conv1D

• ❌ 1DConv

Explanation:
Correct Keras layer:

---

6. Question 6

CSV file has a header you want to skip — what do you call?

• ❌ reader.next

• ✅ next(reader)

• ❌ reader.ignore_header()

• ❌ reader.read(next)

Explanation:
next(reader) reads and discards the first line (header).

---

7. Question 7

Best neural network type for predicting time series like sunspots?

• ❌ RNN/LSTM

• ❌ Convolutions

• ❌ DNN

• ✅ A combination of all other answers

Explanation:
The best model blends:
✔ convolution for pattern extraction
✔ LSTM/RNN for temporal relationships
✔ dense layers for final prediction

---

🧾 Summary Table

---