Run Models
Outputs
Each batch prediction represents a particular run of your selected predictive query, resulting in the creation of a table with the entity primary key IDs (i.e., the remaining IDs after applying entity filters) and the actual prediction output.
The actual prediction output will vary depending on what task types are used in your predictions, and whether they are temporal in nature. For example, if your predictive query is a temporal task type (e.g., temporal regression, temporal binary/multiclass classification or ranking, temporal link prediction), the output table will also include the anchor time of the prediction (i.e., what a prediction horizon start time of “zero” refers to), as configured in your individual batch prediction job; if no anchor time was specified, Kumo by default will use the latest timestamp in your predictive query target formula’s fact table.
Whenever you run a batch prediction, Kumo will automatically first refresh all of the data in your graph to make sure you are using the latest available data (unless the data was already recently refreshed).
You will be able to check what the output looks like under the Output table schema tab.
The
The
The
The
Please refer to the predictive query section to view a full example of a temporal link prediction’s task’s PQL statement, data model, and output.
Please refer to the predictive query section to view a full example of a temporal link prediction’s task’s PQL statement, data model, and output.
Example Prediction Output: Static (Non-Temporal) Binary Classification
Consider the following prediction output for whether a customer will like a particular item on your website:| ENTITY | SCORE |
|---|---|
| 0 | 0.700517 |
| 1 | 0.601757 |
| 2 | 0.446016 |
ENTITY column contains the primary key IDs of another table containing the binary (i.e., true/false) values signifying whether a particular customer liked an item or not—this other table would also include the respective item id and customer id. The SCORE column represents the likelihood that the predictive query would evaluate to true.
Please refer to the predictive query section to view a full example of a non-temporal binary classification task’s PQL statement, data model, and output.
Example Prediction Output: Temporal Binary Classification
If Oct. 17th 2022 is the last date in your data source, and you are predicting whether a customer churns tomorrow (Oct. 18th, 2022) for each of your 10,000 customers, the prediction output table will look like the following:| ENTITY | TIMESTAMP | TARGET_PRED | False_PROB | True_PROB |
|---|---|---|---|---|
| 0 | 2022-10-18 | False | 0.700517 | 0.299483 |
| 1 | 2022-10-18 | False | 0.601757 | 0.398243 |
| 2 | 2022-10-18 | True | 0.446016 | 0.553984 |
False_PROB and True_PROB columns contain the predicted probability this customer churns. The threshold for binary classification you set during batch prediction determines how high the predicted probability must be for us to consider this a positive class prediction. Based on the predicted probability and the threshold, Kumo outputs True (churn) or False (doesn’t churn) in the TARGET_PRED column.
Please refer to the predictive query section to view a full example of a temporal binary classification task’s PQL statement, data model, and output.
Example Prediction Output: Static (Non-Temporal) Regression
If you are predicting the age of your customers—a regression task—the batch prediction output will look like the following:| ENTITY | TARGET_PRED |
|---|---|
| 0 | 29 |
| 1 | 17 |
| 2 | 40 |
TARGET_PRED contains the predicted age of your customer.
Please refer to the predictive query section to view a full example of a non-temporal regression task’s PQL statement, data model, and output.
Example Prediction Output: Temporal Regression
If you would like to predict the amount of money a customer would spend on Oct. 18th—a temporal regression task—the batch prediction output will look like the following:| ENTITY | TIMESTAMP | TARGET_PRED |
|---|---|---|
| 0 | 2024-10-18 | 20.398209 |
| 1 | 2024-10-18 | 5.238759 |
| 2 | 2024-10-18 | 0.000000 |
TARGET_PRED contains the amount of the money the customer is predicted to spend on October 18, 2024.
Please refer to the predictive query section to view a full example of a temporal regression task’s PQL statement, data model, and output.
Example Prediction Output: Static (Non-Temporal) Link Prediction
If you would like to predict the top ten unique items each customer is likely to purchase—a static link prediction task—the batch prediction output will look like the following:| ENTITY | CLASS | SCORE |
|---|---|---|
| 1 | 733749001 | 0.6874649124 |
| 1 | 534746502 | 0.6624874491 |
| 1 | 733001749 | 0.6746829124 |
| 1 | 800441502 | 0.6846829493 |
| 1 | 534746492 | 0.6874624491 |
| 1 | 464953472 | 0.6874491236 |
| 1 | 850244001 | 0.6875469162 |
| 1 | 759871002 | 0.6834917462 |
| 1 | 46494643 | 0.6874624491 |
| 1 | 746534492 | 0.8746749124 |
Example Prediction Output: Temporal Link Prediction
If you would like to predict the unique items each customer is likely to purchase in the next 30 days—a temporal link prediction task, the batch prediction output will look like:| ENTITY | CLASS | SCORE | TIMESTAMP |
|---|---|---|---|
| 0 | 733749001 | 0.6846829493 | 2022-10-18 |
| 1 | 850244001 | 0.6874624491 | 2022-10-18 |
| 2 | 759871002 | 0.6624874491 | 2022-10-18 |