Zoho Sheet STEYX Function: Calculate Regression Standard Error

STEYX(y_values; x_values)

There are two required inputs and no optional arguments/default values documented in this reference. Order matters in interpretation: y comes first. Zoho’s remarks say text, logical values and empty cells are ignored; it lists #N/A! for different-sized x/y inputs and #DIV/0! when either dataset has fewer than three values. Exact behaviour with paired missing cells, imported text or zero-variance x-values was not tested in the native application. Zoho STEYX reference, checked 2026-10-12.

Create a new worksheet for the example and enter these six observations. The A column is the independent variable (week number); B contains the dependent variable (orders). Enter numbers, not quoted text.

This is an illustrative synthetic dataset, not downloaded Zoho Sheet output or a claim about actual business demand. The arithmetic below has been independently checked against these values. Formulas use the semicolon separators shown in Zoho’s English-language function help. If your spreadsheet’s formula parser requires another separator, check the workbook’s locale/settings rather than copying translated notation blindly.

1. Input the dataset in A1:B7 exactly as shown.

2. In D2, enter:

   =STEYX(B2:B7;A2:A7)
   

3. The independently calculated expected result is 1.214985793. With two decimal places it appears as 1.21 orders. It is a measure of historical regression scatter, not a prediction of exactly 1.21 future mistakes.

4. Do not switch the ranges. STEYX(A2:A7;B2:B7) would treat week number as the response variable, returning an error scale measured in weeks instead of orders.

The example’s six numeric paired points exceed the three-point minimum in the official error guidance. Using a single observation or only two observations is not sufficient for the conventional residual standard-error divisor n−2.

We can verify the calculation transparently without running Zoho Sheet. The fitted line for these six points is:

Predicted orders = 1.666666667 + 1.285714286 × Week

That line’s coefficients are obtained from the independently checked sample or, once natively tested, the documented SLOPE and INTERCEPT functions. Calculate each fitted value and residual observed − fitted:

Squaring and summing the unrounded residuals gives a residual sum of squares (SSE) of 5.904761905. There are six pairs, and ordinary simple linear regression estimates two coefficients, leaving n−2 = 4 residual degrees of freedom:

STEYX = SQRT(SSE / (n−2))
      = SQRT(5.90476190476 / 4)
      = 1.21498579259

This is an independently reasoned mathematical calculation, not a native Zoho screenshot or test result. When constructing a cross-check worksheet, place predicted values in C2:C7 and residuals in D2:D7. Example for C2 and D2:

=INTERCEPT($B$2:$B$7;$A$2:$A$7)+SLOPE($B$2:$B$7;$A$2:$A$7)*A2
=B2-C2

Copy both down through row 7. A manual check can square each D2:D7 residual and sum; use full cell precision, not rounded display values, or the last digits will differ. One compact cross-check, using the separately documented SUMPRODUCT, COUNT and SQRT functions, is:

=SQRT(SUMPRODUCT(D2:D7;D2:D7)/(COUNT(B2:B7)-2))

Its expected mathematical result is again 1.214985793. The behaviour of this multi-function composition in Zoho Sheet was not natively verified.

Combine a week-seven expected value with a residual-scale description:

=FORECAST(7;$B$2:$B$7;$A$2:$A$7)
=STEYX($B$2:$B$7;$A$2:$A$7)

The independently calculated expected figures are 10.666666667 forecast orders and 1.214985793 historical residual standard error. A manager might summarize that as “the straight-line forecast is approximately 10.67 orders; the six historical points have a residual scale of about 1.21 orders.” That communicates how the line performed on this sample without claiming that every future value is within ±1.21 units.

A second useful diagnostic is a relative residual scale, calculated solely as a business heuristic:

=STEYX(B2:B7;A2:A7)/AVERAGE(B2:B7)

With the sample mean 6.166666667, the expected ratio is about 0.19702 (19.70%). This is not a standard statistical confidence level, a mean absolute percentage error, or a guarantee about sales variability. It is only a dimensionless way to discuss residual scatter relative to average historical magnitude in a positive-valued series. If the average is zero or near zero, this ratio is unsuitable.

Importantly, you cannot derive a 95% future observation interval merely by subtracting and adding 1.96×STEYX to FORECAST. Prediction intervals depend on sample size, x-distance from the training mean, and an appropriate t critical value, among other assumptions. With only six observations, visually review residuals and consider collecting more data before using any interval for purchasing or staffing.

The first five named errors and the documented size rule come from Zoho STEYX, checked 2026-10-12. No native tests established additional exact error codes or details of blank handling.

Microsoft Excel documents STEYX(known_y's, known_x's) and explicitly includes zeros in referenced arrays while ignoring text, logical values and blanks in arrays/ranges. Its documentation also notes special handling of logical/text-number literals typed directly as arguments—a nuance not established from Zoho’s brief reference. Therefore this guide does not claim that mixed-value handling is identical even though the ordinary numerical regression formula is mathematically equivalent. Excel uses comma-separated syntax in its English reference; Zoho’s English reference shows semicolons. Microsoft STEYX and Zoho STEYX, checked 2026-10-12.

For different analytical questions, use SLOPE for the change in y per unit x, INTERCEPT for the fitted line’s starting intercept, FORECAST for the expected y at a target x, PEARSON for unitless linear correlation, or RSQ for fitted simple-line explained variance in this sample. These are official Zoho Help links checked 2026-10-12.

The official Zoho function page describes no historical first-support version, no per-account feature matrix and no complete cross-platform test matrix. This batch did not establish exact behaviour in Android, iOS, offline clients, imported Excel files or other editions. Workbook locale controls are separately documented at Zoho’s locale settings page, checked 2026-10-12. No claim about other spreadsheet applications beyond the cited Excel detail is warranted.

• Official Zoho documentation checked (2026-10-12): STEYX function, and linked SLOPE, INTERCEPT, FORECAST, RSQ, and PEARSON.
• Other software’s official documentation checked (2026-10-12): Microsoft STEYX. This is compatibility research, not evidence of how Zoho executes a formula.
• Independent calculation: Fitted predictions, residuals, SSE, residual degrees of freedom, regression standard error, and simple ratio checked mathematically against the example dataset.
• Native Zoho Sheet testing: NOT EXECUTED. Native error output, spill behaviour (not applicable to this scalar formula), locale-specific punctuation, mixed-cell handling and all platform editions remain unverified.
• Screenshots: None included; no mock or falsely attributed Zoho Sheet images are used.