Before making Time-Series forecasts or finding impacts, take a moment to familiarize yourself with the high-level process.
We’ve worked hard to keep the high-level process simple. Here’s the basic process:
To generate time-series based forecasts, datasets must have a timestamp column. This is slightly different from regression which does not require a timestamp column, but still might have one.
For example, you may have a data set of house prices with features describing the houses including the year that house was built. Even though you have a date as a feature, this is not a time series problem and you would want to choose to use regression. In time series forecasting, we are generally interested in predicting something that is changing over time, but in this data set, we have several different houses with one date and will be predicting prices of other houses. So, this is a regression problem.
In a time series problem, we expect observations close to each other in time to be more similar than observations far away, after accounting for seasonality. For example, the weather today is usually more similar to the weather tomorrow than the weather a month from now. So, predicting the weather based on past weather observations is a time series problem.
If you’re still not sure which to use, and you have a date/timestamp with target values over time, you can always use our API to try both and compare the results.
Read Sending Data for the technical details.
A Session is simply the a discovery process using the supplied Dataset.
There are two types of time-series based sessions today:
This is where the data science happens at scale. Behind the scenes a host of algorithms will work to discover what makes your Dataset tick, attempting to find what factors are influential to others, where the correlations are and ultimately provide predictions or impact.
Read Sessions for the technical details.
Once the session has completed successfully, the results can be retrieved. Depending on the type of session, the results will contain prediction or impact analysis results.
Read Retrieving a Session for more technical details.