Time series with RNN in TensorFlow - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is a time series in machine learning?

A time series is a sequence of data points collected or recorded at regular time intervals. It shows how something changes over time, like daily temperatures or stock prices.

beginner

Why do we use RNNs for time series data?

RNNs (Recurrent Neural Networks) are good for time series because they remember information from previous steps, helping the model understand patterns over time.

intermediate

What does the 'state' in an RNN represent?

The 'state' in an RNN holds information from previous time steps. It acts like memory, helping the network keep track of what it learned before to make better predictions.

beginner

What is the purpose of splitting time series data into training and testing sets?

Splitting helps us train the model on past data and test it on unseen future data to check if the model can predict well on new time points.

intermediate

How do you evaluate the performance of an RNN on time series forecasting?

Common metrics include Mean Squared Error (MSE) or Mean Absolute Error (MAE), which measure how close the model's predictions are to the actual values.

What type of neural network is best suited for time series data?

A Recurrent Neural Network (RNN)

B Convolutional Neural Network (CNN)

C Feedforward Neural Network

D Generative Adversarial Network (GAN)

RNNs are designed to handle sequences and remember past information, making them ideal for time series.

In time series forecasting, why is it important to keep the order of data?

A Because it helps in data normalization

B Because shuffling improves model accuracy

C Because order does not matter in time series

D Because time order carries important information about trends and patterns

The sequence order is crucial in time series as it reflects how data evolves over time.

Which metric measures the average squared difference between predicted and actual values?

A Mean Absolute Error (MAE)

B Mean Squared Error (MSE)

C Accuracy

D Precision

MSE calculates the average of squared differences, penalizing larger errors more.

What does the 'state' in an RNN help the model do?

A Remember information from previous time steps

B Forget previous inputs

C Increase the size of the input data

D Normalize the data

The state stores past information to help the RNN understand sequences better.

When preparing time series data for RNN, what is a common step?

A Convert data into images

B Randomly shuffle all data points

C Split data into sequences of fixed length

D Remove all time stamps

Splitting data into fixed-length sequences helps the RNN learn patterns over time.

Explain how an RNN processes time series data step-by-step.

Think about how memory helps in understanding sequences.

Describe how you would prepare and evaluate a time series forecasting model using RNN.

Consider data preparation, model training, and checking results.