Decide if statistics such as those found here when going to the "Statistics" tab
If we want to include these as well, modify the ultimatetennisstatistics scrapper to include this information.
A thing to consider is that these statistics are current and should only be used to predict current and future matches, when training a model on historical data we should not feed it this data since it would technically include future data. For example:
If we train on matched until 2015 and test on data from 2016, if we would use statistics such as Ace% that include data till the present it could leak information about future performance of players that would unfairly help the model predict accurately.
Decide if statistics such as those found here when going to the "Statistics" tab
If we want to include these as well, modify the ultimatetennisstatistics scrapper to include this information.
A thing to consider is that these statistics are current and should only be used to predict current and future matches, when training a model on historical data we should not feed it this data since it would technically include future data. For example: If we train on matched until 2015 and test on data from 2016, if we would use statistics such as Ace% that include data till the present it could leak information about future performance of players that would unfairly help the model predict accurately.