Detailed explanation of the steps of implementing multiple linear regression simulation curve algorithm in PHP

This time I will bring you a detailed explanation of the steps of implementing the multiple linear regression simulation curve algorithm in PHP. What are the precautions for implementing the multiple linear regression simulation curve algorithm in PHP. Here are practical cases, let’s take a look.

Multiple linear regression model: y = b1x1 b2x2 b3x3... bnxn;

We are based on a set of data: similar arr_x = [[ 1, 2, 3, 4, 5], [6, 7, 8, 9, 10], [11, 12, 13, 14, 15]]; arr_y = [5, 10, 15]; us The last thing required is an array, including from b1 to bn;

Method: Use the least squares method

Formula:We only use the first half of the formula, that is, the matrix To calculate

X in the formula is arr_x, two-dimensional arrayWe can think of it as a matrix, y in the formula is arr_y, and also think of it as a matrix (5 , 10, 15) , but it should be written vertically.

Then according to the formula, we will find that we need to use matrix multiplication, transposition, and inversion; so the following codes are given one by one:

public function get_complement($data, $i, $j) {
  /* x和y为矩阵data的行数和列数 */
  $x = count($data);
  $y = count($data[0]);
  /* data2为所求剩余矩阵 */
  $data2 =[];
  for ($k = 0; $k < $x -1; $k++) {
    if ($k < $i) {
      for ($kk = 0; $kk < $y -1; $kk++) {
        if ($kk < $j) {
          $data2[$k][$kk] = $data[$k][$kk];
        } else {
          $data2[$k][$kk] = $data[$k][$kk +1];
        }
      }
    } else {
      for ($kk = 0; $kk < $y -1; $kk++) {
        if ($kk < $j) {
          $data2[$k][$kk] = $data[$k +1][$kk];
        } else {
          $data2[$k][$kk] = $data[$k +1][$kk +1];
        }
      }
    }
  }
  return $data2;
}
/* 计算矩阵行列式 */
public function cal_det($data) {
  $ans = 0;
  if (count($data[0]) === 2) {
    $ans = $data[0][0] * $data[1][1] - $data[0][1] * $data[1][0];
  } else {
    for ($i = 0; $i < count($data[0]); $i++) {
      $data_temp = $this->get_complement($data, 0, $i);
      if ($i % 2 === 0) {
        $ans = $ans + $data[0][$i] * ($this->cal_det($data_temp));
      } else {
        $ans = $ans - $data[0][$i] * ($this->cal_det($data_temp));
      }
    }
  }
  return $ans;
}
/*计算矩阵的伴随矩阵*/
public function ajoint($data) {
  $m = count($data);
  $n = count($data[0]);
  $data2 =[];
  for ($i = 0; $i < $m; $i++) {
    for ($j = 0; $j < $n; $j++) {
      if (($i + $j) % 2 === 0) {
        $data2[$i][$j] = $this->cal_det($this->get_complement($data, $i, $j));
      } else {
        $data2[$i][$j] = - $this->cal_det($this->get_complement($data, $i, $j));
      }
    }
  }
  return $this->trans($data2);
}
/*转置矩阵*/
public function trans($data) {
  $i = count($data);
  $j = count($data[0]);
  $data2 =[];
  for ($k2 = 0; $k2 < $j; $k2++) {
    for ($k1 = 0; $k1 < $i; $k1++) {
      $data2[$k2][$k1] = $data[$k1][$k2];
    }
  }
  /*将矩阵转置便可得到伴随矩阵*/
  return $data2;
}
/*求矩阵的逆，输入参数为原矩阵*/
public function inv($data) {
  $m = count($data);
  $n = count($data[0]);
  $data2 =[];
  $det_val = $this->cal_det($data);
  $data2 = $this->ajoint($data);
  for ($i = 0; $i < $m; $i++) {
    for ($j = 0; $j < $n; $j++) {
      $data2[$i][$j] = $data2[$i][$j] / $det_val;
    }
  }
  return $data2;
}
/*求两矩阵的乘积*/
public function getProduct($data1, $data2) {
  /*$data1 为左乘矩阵*/
  $m1 = count($data1);
  $n1 = count($data1[0]);
  $m2 = count($data2);
  $n2 = count($data2[0]);
  $data_new =[];
  if ($n1 !== $m2) {
    return false;
  } else {
    for ($i = 0; $i <= $m1 -1; $i++) {
      for ($k = 0; $k <= $n2 -1; $k++) {
        $data_new[$i][$k] = 0;
        for ($j = 0; $j <= $n1 -1; $j++) {
          $data_new[$i][$k] += $data1[$i][$j] * $data2[$j][$k];
        }
      }
    }
  }
  return $data_new;
}
/*多元线性方程*/
public function getParams($arr_x, $arr_y) {
  $final =[];
  $arr_x_t = $this->trans($arr_x);
  $result = $this->getProduct($this->getProduct($this->inv($this->getProduct($arr_x_t, $arr_x)), $arr_x_t), $arr_y);
  foreach ($result as $key => $val) {
    foreach ($val as $_k => $_v) {
      $final[] = $_v;
    }
  }
  return $final;
}

The last getParams( ) method is the final method to find the b parameter array, just pass in a two-dimensional array arr_x, and a one-dimensional array arr_y.

I believe you have mastered the method after reading the case in this article. For more exciting information, please pay attention to other related articles on the php Chinese website!

Recommended reading:

Detailed explanation of the steps to convert an array using foreach in PHP

#php Detailed explanation of the steps to delete the value element in a one-dimensional array

The above is the detailed content of Detailed explanation of the steps of implementing multiple linear regression simulation curve algorithm in PHP. For more information, please follow other related articles on the PHP Chinese website!