What are the differences between lxml and ElementTree?

I wouldn't say that lxml is faster than ET across the board as both modules offer tons of functionality. To provide a little context, ElementTree also supports XPath, but particularly ET has a unique and useful function called iterparse() that remakes the XML document as an iterable. This results in much faster parsing, especially for large XML files.

The ET API itself creates Element types which are a hybrid cross between a list and dictionary. This can mean headaches for those new to the module, but sit down with it and you'll see that it's pretty flexible.

ElementTree comes built-in with the Python standard library which includes other data modules types such as json and csv. This means the module ships with each installation of Python. For most normal XML operations including building document trees and simple searching and parsing of element attributes and node values, even namespaces, ElementTree is a reliable handler.

Lxml is a third-party module that requires installation. In many ways lxml actually extends ElementTree as most operations in the built-in module are available. Chief among this extension is that lxml supports both XPath 1.0 and XSLT 1.0. Additionally, lxml can parse HTML documents that are not XML compliant and hence is used for web-scraping operations and even as the parser in BeautifulSoup and engine in Pandas, pandas.read_html(). Other useful, common features of lxml include pretty_print output, objectify, and sax support. Of course too as a third-party module, versions with additional features are readily accessible compared to the standard library.