Skip to content

Overview

The following subsections go through all the available outputs from TIM (RT)InstantML Forecasting. Each subsection is comprised of a label specifying if it is related to InstantML or RTInstantML, a text description, and TIM API notation. To learn more about the architectural differences between the InstantML and RTInstantML TIM Forecasting see this overview.

Below is a table with all the available outputs from TIM (RT)InstantML Forecasting.

Configuration field RTInstantML InstantML model building InstantML forecasting
Difficulty x x
Simple importances x x
Extended importances x x
Data offsets x
Prediction x x x
Prediction intervals x
Low prediction quality datetimes x x x
Aggregated predictions x x x
Raw predictions x x
Result explanations x x x

Difficulty

Relates to InstantML model building phase and RTInstantML

This is a simple measure of how difficult it should be to model given data. It ranges from 0 to 100 percent and is calculated as 1 minus ratio of explained variance to original variance when using a simple regression model. Completely random data will have difficulty close to 100 and vice versa.

"dataDifficulty": 37.5

Simple importances

Relates to InstantML model building phase and RTInstantML

A measure of how much the individual predictors entering the model building contribute to the model and whether it is worth to collect them. The individual importances sum up to 100 percent.

"simpleImportances": [
    {
        "predictorName": "Water_pressure",
        "importance": 45.0
    },
    {
        ...
    }
 ]

Extended importances

Relates to InstantML model building phase and RTInstantML

A more complicated measure not related to the original predictors, but model features - transformations of predictors done by TIM during the model building phase. It is also done separately for every specific time of the day if the model is time specific. On top of that, if the ModelZOO consists of several models able to model one specific time of the day, the importance is aggregated across them. It sums up to 100 for every part of the day. Before the aggregation across several models happens, the measure represents a portion of variance explained by this transformed predictor. Extended importances can be used for creating nice tree maps.

"extendedImportances": [
    {
        "termName": "Load(t-31) & #MA(t-13, w:2)",
        "importance": 10.35,
        "time": "00:00:00",
        "type": "Interaction"
    },
    {
        ...
    }
]

Data offsets

Relates to InstantML model building phase

To avoid always plugging all data to TIM when making predictions, data offsets exist to tell you how much of history is really needed. For each of the original predictors you get an integer value that tells you how far in the history the model potentially looks (in number of samples) when predicting a specific timestamp. This way you know what your model needs and you do not have to transfer unnecessary loads of data when predicting with TIM.

"dataOffsets": [
    {
        "uniqueName": "STEAM",
        "from": {
            "baseUnit": "Sample",
            "offset": -430
        }
    }
]

Prediction

Relates to InstantML model building phase, forecasting phase and RTInstantML

Timestamps with prediction for the desired period.

"prediction": {
    "values": {
        "2004-10-30T00:00:00Z": 15963.314,
        "2004-10-31T00:00:00Z": 15261.849,
        "2004-11-01T00:00:00Z": 15753.487
    }
}

Prediction intervals

Relates to RTInstantML

Timestamps with lower and upper values of the prediction intervals for the desired period.

"prediction": {
    "predictionIntervals": {
        "lowerValues": {
            "2004-10-30T00:00:00Z": 15725.223,
            "2004-10-31T00:00:00Z": 15110.018,
            "2004-11-01T00:00:00Z": 15385.933
        },
        "upperValues": {
            "2004-10-30T00:00:00Z": 16050.868,
            "2004-10-31T00:00:00Z": 15422.564,
            "2004-11-01T00:00:00Z": 15991.357
        }
    }
}

Low prediction quality datetimes

Relates to InstantML model building phase, forecasting phase and RTInstantML

Sometimes ModelZOO doesn't contain models that give a reasonable prediction. This might be caused by providing not enough data, or missing values. In this case the set of dummy models is used. Quality of their predictions tends to be lower.

"prediction": {
    "lowPredictionQualityDateTimes": [
        "2004-09-02T00:00:00Z",
        "2004-09-03T00:00:00Z"
    ]
}

Aggregated predictions

Relates to InstantML model building phase, forecasting phase and RTInstantML

Aggregated prediction output, see Raw and aggregated predictions section to learn more.

"aggregatedPredictions": [
    {
        "day": 2,
        "predictionTime": "07:00:00",
        "values": {
            "2018-11-01T00:00:00.000Z": -0.548916,
            "2018-11-01T01:00:00.000Z": -0.557343,
            "2018-11-01T02:00:00.000Z": -0.553979
        }
    {
        ...
    }
]

Raw predictions

Relates to InstantML model building phase and forecasting phase

Raw prediction output, see Raw and aggregated predictions section to learn more.

"rawPredictions": [
    {
        "predictionDateTime": "2018-10-30T07:00:00.000Z",
        "values": {
            "2018-11-01T00:00:00.000Z": -0.548916,
            "2018-11-01T01:00:00.000Z": -0.557343,
            "2018-11-01T02:00:00.000Z": -0.553979
        }
    {
        ...
    }
]

Result explanations

Relates to InstantML model building phase, forecasting phase and RTInstantML

Error and warning messages that can appear after running an (RT)InstantML forecast. The full list can be found here.

"resultExplanations": [
    {
        "index": 1,
        "message": "Predictor name "humidity" is not unique!"
    }
]