Often simplicity goes unnoticed – we overlook it, take it for granted just because it feels so simple. But ‘simple’ processes can be very complex in nature. At BigML we don’t ask any question about the data you are feeding us. It is super simple to just click and drag a file to the website or link to a remote source. “Just give us your data and we’ll make sense of it” requires quite a bit of complexity. Yet we think it is a vital part of making machine learning simple. Now we have added some more logic: automatic date parsing.
As of now, whenever we detect a date/time field in your source, we’ll automatically add a few fields. We’ll split the field into its respective parts (year, month, day, hour, minute, second and even millisecond) and add day of the week. This way your analysis can take into account that transactions peak on certain days of the week, periods of the year or even around specific times of the day. And it’s all done automatically. The one restriction we are applying at the moment is that your date(/time) conforms to ISO 8601. But hey, if you have a bunch of data that contains a different date type, let us know so we can look into it! ISO 8601 format is just a start.
As always, we are committed to making machine learning technology as simple as possible. Enjoy it!