UPDATE: Implements CORREL, SLOPE, INTERCEPT, RSQ and STEYX

These are all functions that follow a very simmilar path code

base/src/expressions/parser/static_analysis.rs

@@ -990,6 +990,11 @@ fn get_function_args_signature(kind: &Function, arg_count: usize) -> Vec<Signatu
         Function::Sumx2my2 => vec![Signature::Vector; 2],
         Function::Sumx2py2 => vec![Signature::Vector; 2],
         Function::Sumxmy2 => vec![Signature::Vector; 2],
+        Function::Correl => vec![Signature::Vector; 2],
+        Function::Rsq => vec![Signature::Vector; 2],
+        Function::Intercept => vec![Signature::Vector; 2],
+        Function::Slope => vec![Signature::Vector; 2],
+        Function::Steyx => vec![Signature::Vector; 2],
     }
 }
 
@@ -1324,5 +1329,10 @@ fn static_analysis_on_function(kind: &Function, args: &[Node]) -> StaticResult {
         Function::Sumx2my2 => StaticResult::Scalar,
         Function::Sumx2py2 => StaticResult::Scalar,
         Function::Sumxmy2 => StaticResult::Scalar,
+        Function::Correl => StaticResult::Scalar,
+        Function::Rsq => StaticResult::Scalar,
+        Function::Intercept => StaticResult::Scalar,
+        Function::Slope => StaticResult::Scalar,
+        Function::Steyx => StaticResult::Scalar,
     }
 }

base/src/functions/mod.rs

@@ -421,10 +421,16 @@ pub enum Function {
     Dvar,
     Dvarp,
     Dstdevp,
+
+    Correl,
+    Rsq,
+    Intercept,
+    Slope,
+    Steyx,
 }
 
 impl Function {
-    pub fn into_iter() -> IntoIter<Function, 328> {
+    pub fn into_iter() -> IntoIter<Function, 333> {
         [
             Function::And,
             Function::False,
@@ -754,6 +760,11 @@ impl Function {
             Function::VarA,
             Function::WeibullDist,
             Function::ZTest,
+            Function::Correl,
+            Function::Rsq,
+            Function::Intercept,
+            Function::Slope,
+            Function::Steyx,
         ]
         .into_iter()
     }
@@ -1234,6 +1245,11 @@ impl Function {
             "SUMX2MY2" => Some(Function::Sumx2my2),
             "SUMX2PY2" => Some(Function::Sumx2py2),
             "SUMXMY2" => Some(Function::Sumxmy2),
+            "CORREL" => Some(Function::Correl),
+            "RSQ" => Some(Function::Rsq),
+            "INTERCEPT" => Some(Function::Intercept),
+            "SLOPE" => Some(Function::Slope),
+            "STEYX" => Some(Function::Steyx),
 
             _ => None,
         }
@@ -1573,6 +1589,11 @@ impl fmt::Display for Function {
             Function::Sumx2my2 => write!(f, "SUMX2MY2"),
             Function::Sumx2py2 => write!(f, "SUMX2PY2"),
             Function::Sumxmy2 => write!(f, "SUMXMY2"),
+            Function::Correl => write!(f, "CORREL"),
+            Function::Rsq => write!(f, "RSQ"),
+            Function::Intercept => write!(f, "INTERCEPT"),
+            Function::Slope => write!(f, "SLOPE"),
+            Function::Steyx => write!(f, "STEYX"),
         }
     }
 }
@@ -1929,6 +1950,11 @@ impl Model {
             Function::Sumx2my2 => self.fn_sumx2my2(args, cell),
             Function::Sumx2py2 => self.fn_sumx2py2(args, cell),
             Function::Sumxmy2 => self.fn_sumxmy2(args, cell),
+            Function::Correl => self.fn_correl(args, cell),
+            Function::Rsq => self.fn_rsq(args, cell),
+            Function::Intercept => self.fn_intercept(args, cell),
+            Function::Slope => self.fn_slope(args, cell),
+            Function::Steyx => self.fn_steyx(args, cell),
         }
     }
 }

base/src/functions/statistical/correl.rs

@@ -0,0 +1,227 @@
+use crate::expressions::types::CellReferenceIndex;
+use crate::{
+    calc_result::CalcResult, expressions::parser::Node, expressions::token::Error, model::Model,
+};
+
+impl Model {
+    // CORREL(array1, array2) - Returns the correlation coefficient of two data sets
+    pub(crate) fn fn_correl(&mut self, args: &[Node], cell: CellReferenceIndex) -> CalcResult {
+        let (_, _, values_left, values_right) = match self.fn_get_two_matrices(args, cell) {
+            Ok(s) => s,
+            Err(e) => return e,
+        };
+
+        let mut n = 0.0;
+        let mut sum_x = 0.0;
+        let mut sum_y = 0.0;
+        let mut sum_x2 = 0.0;
+        let mut sum_y2 = 0.0;
+        let mut sum_xy = 0.0;
+
+        for (x_opt, y_opt) in values_left.into_iter().zip(values_right.into_iter()) {
+            if let (Some(x), Some(y)) = (x_opt, y_opt) {
+                n += 1.0;
+                sum_x += x;
+                sum_y += y;
+                sum_x2 += x * x;
+                sum_y2 += y * y;
+                sum_xy += x * y;
+            }
+        }
+
+        // Need at least 2 valid pairs
+        if n < 2.0 {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "CORREL requires at least two numeric data points in each range".to_string(),
+            );
+        }
+
+        let num = n * sum_xy - sum_x * sum_y;
+        let denom_x = n * sum_x2 - sum_x * sum_x;
+        let denom_y = n * sum_y2 - sum_y * sum_y;
+        let denom = (denom_x * denom_y).sqrt();
+
+        if denom == 0.0 || !denom.is_finite() {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "Division by zero in CORREL".to_string(),
+            );
+        }
+
+        let r = num / denom;
+        CalcResult::Number(r)
+    }
+
+    // SLOPE(known_y's, known_x's) - Returns the slope of the linear regression line
+    pub(crate) fn fn_slope(&mut self, args: &[Node], cell: CellReferenceIndex) -> CalcResult {
+        let (_rows, _cols, values_y, values_x) = match self.fn_get_two_matrices(args, cell) {
+            Ok(s) => s,
+            Err(e) => return e,
+        };
+
+        let mut n = 0.0;
+        let mut sum_x = 0.0;
+        let mut sum_y = 0.0;
+        let mut sum_x2 = 0.0;
+        let mut sum_xy = 0.0;
+
+        let len = values_y.len().min(values_x.len());
+        for i in 0..len {
+            if let (Some(y), Some(x)) = (values_y[i], values_x[i]) {
+                n += 1.0;
+                sum_x += x;
+                sum_y += y;
+                sum_x2 += x * x;
+                sum_xy += x * y;
+            }
+        }
+
+        if n < 2.0 {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "SLOPE requires at least two numeric data points".to_string(),
+            );
+        }
+
+        let denom = n * sum_x2 - sum_x * sum_x;
+        if denom == 0.0 || !denom.is_finite() {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "Division by zero in SLOPE".to_string(),
+            );
+        }
+
+        let num = n * sum_xy - sum_x * sum_y;
+        let slope = num / denom;
+
+        CalcResult::Number(slope)
+    }
+
+    // INTERCEPT(known_y's, known_x's) - Returns the y-intercept of the linear regression line
+    pub(crate) fn fn_intercept(&mut self, args: &[Node], cell: CellReferenceIndex) -> CalcResult {
+        let (_rows, _cols, values_y, values_x) = match self.fn_get_two_matrices(args, cell) {
+            Ok(s) => s,
+            Err(e) => return e,
+        };
+
+        let mut n = 0.0;
+        let mut sum_x = 0.0;
+        let mut sum_y = 0.0;
+        let mut sum_x2 = 0.0;
+        let mut sum_xy = 0.0;
+
+        let len = values_y.len().min(values_x.len());
+        for i in 0..len {
+            if let (Some(y), Some(x)) = (values_y[i], values_x[i]) {
+                n += 1.0;
+                sum_x += x;
+                sum_y += y;
+                sum_x2 += x * x;
+                sum_xy += x * y;
+            }
+        }
+
+        if n < 2.0 {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "INTERCEPT requires at least two numeric data points".to_string(),
+            );
+        }
+
+        let denom = n * sum_x2 - sum_x * sum_x;
+        if denom == 0.0 || !denom.is_finite() {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "Division by zero in INTERCEPT".to_string(),
+            );
+        }
+
+        let num = n * sum_xy - sum_x * sum_y;
+        let slope = num / denom;
+        let intercept = (sum_y - slope * sum_x) / n;
+
+        CalcResult::Number(intercept)
+    }
+
+    // STEYX(known_y's, known_x's) - Returns the standard error of the predicted y-values
+    pub(crate) fn fn_steyx(&mut self, args: &[Node], cell: CellReferenceIndex) -> CalcResult {
+        let (_rows, _cols, values_y, values_x) = match self.fn_get_two_matrices(args, cell) {
+            Ok(s) => s,
+            Err(e) => return e,
+        };
+
+        let mut n = 0.0;
+        let mut sum_x = 0.0;
+        let mut sum_y = 0.0;
+        let mut sum_x2 = 0.0;
+        let mut sum_xy = 0.0;
+
+        // We need the actual pairs again later for residuals
+        let mut pairs: Vec<(f64, f64)> = Vec::new();
+
+        let len = values_y.len().min(values_x.len());
+        for i in 0..len {
+            if let (Some(y), Some(x)) = (values_y[i], values_x[i]) {
+                n += 1.0;
+                sum_x += x;
+                sum_y += y;
+                sum_x2 += x * x;
+                sum_xy += x * y;
+                pairs.push((x, y));
+            }
+        }
+
+        // Need at least 3 points for STEYX (n - 2 in denominator)
+        if n < 3.0 {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "STEYX requires at least three numeric data points".to_string(),
+            );
+        }
+
+        let denom = n * sum_x2 - sum_x * sum_x;
+        if denom == 0.0 || !denom.is_finite() {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "Division by zero in STEYX".to_string(),
+            );
+        }
+
+        let num = n * sum_xy - sum_x * sum_y;
+        let slope = num / denom;
+        let intercept = (sum_y - slope * sum_x) / n;
+
+        // Sum of squared residuals: Σ (y - ŷ)^2, ŷ = intercept + slope * x
+        let mut sse = 0.0;
+        for (x, y) in pairs {
+            let y_hat = intercept + slope * x;
+            let diff = y - y_hat;
+            sse += diff * diff;
+        }
+
+        let dof = n - 2.0;
+        if dof <= 0.0 {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "STEYX has non-positive degrees of freedom".to_string(),
+            );
+        }
+
+        let sey = (sse / dof).sqrt();
+        if !sey.is_finite() {
+            return CalcResult::new_error(Error::DIV, cell, "Numerical error in STEYX".to_string());
+        }
+
+        CalcResult::Number(sey)
+    }
+}

base/src/functions/statistical/mod.rs

@@ -1,6 +1,7 @@
 mod beta;
 mod binom;
 mod chisq;
+mod correl;
 mod count_and_average;
 mod covariance;
 mod devsq;

base/src/functions/statistical/pearson.rs

@@ -63,4 +63,51 @@ impl Model {
 
         CalcResult::Number(num / denom)
     }
+
+    // RSQ(array1, array2) = CORREL(array1, array2)^2
+    pub(crate) fn fn_rsq(&mut self, args: &[Node], cell: CellReferenceIndex) -> CalcResult {
+        let (_rows, _cols, values1, values2) = match self.fn_get_two_matrices(args, cell) {
+            Ok(s) => s,
+            Err(e) => return e,
+        };
+
+        let mut n = 0.0_f64;
+        let mut sum_x = 0.0_f64;
+        let mut sum_y = 0.0_f64;
+        let mut sum_x2 = 0.0_f64;
+        let mut sum_y2 = 0.0_f64;
+        let mut sum_xy = 0.0_f64;
+
+        let len = values1.len().min(values2.len());
+        for i in 0..len {
+            if let (Some(x), Some(y)) = (values1[i], values2[i]) {
+                n += 1.0;
+                sum_x += x;
+                sum_y += y;
+                sum_x2 += x * x;
+                sum_y2 += y * y;
+                sum_xy += x * y;
+            }
+        }
+
+        if n < 2.0 {
+            return CalcResult::new_error(
+                Error::DIV,
+                cell,
+                "RSQ requires at least two numeric data points in each range".to_string(),
+            );
+        }
+
+        let num = n * sum_xy - sum_x * sum_y;
+        let denom_x = n * sum_x2 - sum_x * sum_x;
+        let denom_y = n * sum_y2 - sum_y * sum_y;
+        let denom = (denom_x * denom_y).sqrt();
+
+        if denom == 0.0 || !denom.is_finite() {
+            return CalcResult::new_error(Error::DIV, cell, "Division by zero in RSQ".to_string());
+        }
+
+        let r = num / denom;
+        CalcResult::Number(r * r)
+    }
 }