Depending on how the Predict model fails, Mixpanel will send an email message as well as an in-app alert to you with an explanation why we were unable to calculate the conversion. There are a few different reasons a Predict model will fail.
- Insufficient data to generate model - Not enough data in your project to generate a Predict goal.
- Insufficient positive examples to generate predictive model - Not enough customers have converted on your Predict goal.
- Insufficient negative examples to generate predictive model -Not enough customers have not converted on your Predict goal.
The above messages refer to the amount of data in your project that is used to build the Predict model. It generally means that there are not enough examples in your data, so not enough customers have converted (or not converted) to your end goal, so Predict is unable to build a model.
- Not enough People Profiles set up - Not enough people profiles set up for this project.
You need People Profiles to use Predict, if you do not have enough set up, Mixpanel will be unable to accurately build a Predict model.
- Not enough signal in the data to generate a predictive model - The data is too noisy to accurately build a model.
This message means that the model was unable to find a pattern in your data. Please try selecting a different event or frequency for your model.
- Unknown error - Generally refers to a system error.