This function returns a primary layer and several auxiliary layers. The function can output a secondary (ARTFITMETADATA) result set with goodness-of-fit metadata, and a tertiary (ARTFITRESIDUALS) result set with residuals from the fitting operation.
The function returns a primary RETURNS TABLE, which has an ordered list of solved coefficients. If you specified MODEL_STATS (1), then a secondary layer has the model statistics. If you specified RESIDUALS (1), then a tertiary layer has residual statistics.
You display the primary layer with a SELECT statement. You display the secondary (ARTFITMETADATA) and tertiary (ARTFITRESIDUALS) layers with the TD_EXTRACT_RESULTS function.
Key to Variables in Formulas in RETURNS TABLE Schema Description Columns
| Variable | Description |
|---|---|
| Yactual[i] | Actual coefficient with index (position) i in formula. |
| Ycalc[i] | Calculated coefficient with index (position) i in formula. |
| n | Number of sample points in input time series. |
| R[i] | Variance between actual and calculated coefficients with index (position) i in formula. |
| k | Number of explanatory variables in formula. |
Primary RETURNS TABLE Schema
| Name | Data Type | Description |
|---|---|---|
| derived-series-identifier | Varies | The resultant series identifier inherited from the SERIES_ID. |
| ROW_I | INTEGER | Index (position) of coefficient in formula. |
| COEFF_NAME | VARCHAR(120) | Coefficient name. |
| COEFF_VALUE | FLOAT | Calculated value of coefficient. |
| STD_ERROR | FLOAT | [Appears only with COEFF_STATS (1).] Standard error associated with calculated value of coefficient. |
| TSTAT_VALUE | FLOAT | [Appears only with COEFF_STATS (1).] The significance of coefficient relative to 0.t-statistic calculated for coefficient: coefficient_value / STD_ERROR (coefficient_value) |
| TSTAT_PROB | FLOAT | [Appears only with COEFF_STATS (1).] Probability associated with t-statistic calculated for input time series if coefficient is 0. (More specifically, probability of getting |t| value as large as that calculated for input time series if coefficient is 0.) |
| SIGNIF_RATING | VARCHAR(20) | [Appears only with COEFF_STATS (1).] The rating or confidence level range associated with the coefficient. |
| CONF_INT_LOW | FLOAT | Lowest value of coefficient, given confidence level specified in brackets, such as '[0.95]'. Only valid with COEFF_STATS (1). |
| CONF_INT_HIGH | FLOAT | Highest value of coefficient, given confidence level specified in brackets, such as '[0.95]'. Only valid with COEFF_STATS(1). |
Secondary RETURNS TABLE Schema
| Name | Data Type | Description |
|---|---|---|
| derived-series-identifier | VARCHAR | The resultant series identifier inherited from the SERIES_ID. |
| ROW_I | Integer | Index of coefficient in formula. |
| STD_ERROR2 | FLOAT | Standard error (standard deviation) associated with formula: 1 / (n - 2) * SUM (Yactual[i] - Ycalc[i])2 |
| STD_ERROR_DF | FLOAT | Degrees of freedom associated with standard error calculation. |
| MULTIPLE_R_SQUARED | FLOAT | Fraction of variance. |
| ADJUSTED_R_SQUARED | FLOAT | MULTIPLE_R_SQUARED value adjusted to lessen higher number of explanatory variables. |
| FSTATISTIC | FLOAT | The F-statistic associated with a ratio of the two variances (explained / unexplained) . |
| EXPLAINED_DF | FLOAT | Degrees of freedom associated with explained part of the F-statistic. |
| UNEXPLAINED_DF | FLOAT | Degrees of freedom associated with unexplained part of the F-statistic. |
| PROB_VALUE | FLOAT | Probability associated with the F-statistic. |
Tertiary RETURNS TABLE Schema
| Name | Data Type | Description |
|---|---|---|
| derived-series-identifier | VARCHAR | The resultant series identifier inherited from the SERIES_ID. |
| ROW_I | Integer | Index of coefficient in formula. |
| X1 | FLOAT | First explanatory variable in formula, second FIELD in PAYLOAD. |
| ACTUAL_VALUE | FLOAT | Actual value of response variable. |
| CALC_VALUE | FLOAT | Value of response variable calculated using formula. |
| RESIDUAL | FLOAT | Difference between calculated and actual response variable values. |