Part 3

True or False:  Traditional methods for time series forecasting primarily involve neural networks.

• True

• False

Traditional methods for time series forecasting primarily involve statistical techniques that aim to capture linear relationships within the data and are typically based on assumptions about the underlying data distribution.

• True

• False

True or False: Exponential Smoothing is a deep learning based method.

• True

• False

Select the traditional time series forecasting methods.

• Autoregressive (AR)

• Moving Average (MA)

• Exponential Smoothing

• LSTM

This model predicts future values based on their past values. It assumes that past values have a linear influence on future predictions.

• DNN

• LSTM

• Autoregressive (AR)

• GRU

This method predicts future values based on the residuals (errors) of past predictions.

• Exponential Smoothing

• Moving Average (MA)

• LSTM

• Transformer

Select the adventages of traditional time series forecasting methods.

• Interpetability

• Efficiency

• Complex Patterns

• Manual Feature Engineering

Select the limitations of traditional time series forecasting methods.

• Assumptions: Many methods assume linearity and stationarity in the data.

• Complex Patterns: Struggle with capturing complex, non-linear interdependencies in the data.

• Manual Feature Engineering: Require domain knowledge to craft features and transform data.

• Many traditional methods have been around for decades and are supported by extensive literature.

True or False: Deep Neural Networks are designed for sequential data, capturing long-term dependencies in time series data.

• False

• True

Recurrent Neural Networks are:

• Designed for sequential data, capturing short-term dependencies in time series data.

• Designed for sequential data, capturing long-term dependencies in time series data.

• Designed for 2D data, capturing long-term dependencies in time series data.

• Designed for 3D data, capturing long-term dependencies in time series data.