A Practical Approach to Timeseries Forecasting Using Python - LSTM Forecasting - Project 2: Microsoft Corporation Stock

The video tutorial covers the use of two libraries: the U.S. Federal Holiday calendar and custom business day. It explains how to set up custom business days, define prediction dates, and configure the frequency for predictions. The tutorial demonstrates running model predictions, performing inverse transformations, and creating a forecast data frame. It also includes visualizing results using C bond and discusses exploring different LSTM models for improved outcomes.

Which two libraries are introduced for handling holidays and business days?

• U.S. Federal Holiday calendar and custom business day

• Matplotlib and Seaborn

• Pandas and NumPy

• SciPy and Scikit-learn

What is the purpose of the 'custom business day' in the context of the video?

• To calculate weekends

• To import data from a database

• To visualize data trends

• To define business days excluding U.S. Federal Holidays

How many past days are considered for prediction in the video?

• 10

• 100

• 5

• 50

What is the significance of setting the 'frequency' in the prediction dates?

• To define the data visualization style

• To set the prediction model type

• To specify the number of predictions

• To determine the time interval between predictions

What is the role of 'inverse_transform' in the prediction process?

• To scale the data to a new range

• To convert predictions back to the original scale

• To split the data into training and testing sets

• To visualize the prediction results

What is the format used for dates in the final data frame for forecasting?

• MM/DD/YYYY

• YYYY/DD/MM

• YYYY-MM-DD

• DD-MM-YYYY

Which library is used for visualizing the LSTM results?

• Seaborn

• Bokeh

• Plotly

• Matplotlib

What is the main issue with the single LSTM model as discussed in the video?

• It is not compatible with Seaborn

• It is too complex to implement

• It does not provide satisfactory results

• It requires too much data

What are the alternative LSTM models suggested for better results?

• Feedforward Neural Networks and CNNs

• Simple RNNs and GRUs

• Convolutional LSTMs and RNNs

• Stacked LSTMs and Bi-LSTMs

What is the purpose of using 'PD2 datetime' in the visualization process?

• To convert strings to datetime objects

• To calculate the difference between dates

• To format dates for display

• To filter data by date