Trivago is a German technology company that is essentially a hotel price comparison website. It claims to be the world’s largest online hotel search site. As part of the information on their websiteTrivago displays the Trivago Rating Index (TRI), a number between 0 and 100 for every hotel (with 100 being the highest rating and 0 the lowest).

In the data provided are over 4,500 text comments from Trivago for various hotels and lodges. Also in the data is a variable titled “Score High/Low” that translates to a “1” if the TRI is above 49 and a “0” if the TRI is 49 or below.

Use a sample of 500 of the text comments to attempt to classify a hotel as either a “1” or “0.” After mining the text, you will need to apply a classification model to the SVD-derived concept document matrix. The target for the classification model will be the “Score High/Low” variable.

Be aware that there are few negative reviews, and so the naive model would be to simply rate every hotel as a “1.” Try to estimate a model using the text comments alone that does a better job than a naive model. Use standard diagnostic statistics to evaluate the model (C11P8).