Login

Over time, I've developed a set of machine learning best practices that have served me well in various projects. Some machine learning best practices seem obvious in retrospect, but took me a while to fully appreciate.

One machine learning best practice I now follow religiously is thorough data validation before training. What machine learning best practices have become essential in your work? Any machine learning best practices you think are often overlooked?

I’ve been trying to rely more on the bus and train lately, but I keep getting thrown off by the real-time arrival info just being wrong. The app says five minutes, then it jumps to fifteen, and sometimes the vehicle just never appears. Has anyone else had this happen consistently on a specific route?

Yeah, I’ve noticed the same on my bus route. ETA says five, then it jumps to fifteen and the bus just vanishes from the screen.

I try to give it a margin and still end up waiting longer than planned; I’ve started leaving five minutes earlier than the app suggests.

On another route the times feel more reliable, but on mine it seems completely random.

Do you think the real problem is the data feed quality or just buses bunching up more than the app can predict?

I did file a complaint once and got a canned reply about service disruptions, not much help with the feed issue.

Sometimes the live map shows the bus in the next block and then it never stops at my stop.

I’ve started carrying a book or a podcast because the wait is so variable; it’s easier to cope.

Login
Username:
Password:	Lost Password?
	Remember me