Does machine learning app collect data from servers using UF or we need to write our own data collection logic and feed data into splunk and than we can use ML app ?
The ML app can only analyze whatever data has been indexed into splunk -- or input as csv etc -- and it only does so when you point it at the data.
So, you need to decide what data to collect, collect it however you want (rsyslog or syslogng or some other app), put it into splunk however you want (forwarders, indexers, CONF files set up per your specs) and then run through the various parts of the ML process to point the ML toward answering any questions you might have about the data.
The ML app can only analyze whatever data has been indexed into splunk -- or input as csv etc -- and it only does so when you point it at the data.
So, you need to decide what data to collect, collect it however you want (rsyslog or syslogng or some other app), put it into splunk however you want (forwarders, indexers, CONF files set up per your specs) and then run through the various parts of the ML process to point the ML toward answering any questions you might have about the data.