Added Q: Linear Regression using OLS

questions/186_linear_regression_ordinary_least_squares/description.md

@@ -0,0 +1,15 @@
+### Problem
+
+Implement simple linear regression using Ordinary Least Squares (OLS). Given 1D inputs `X` and targets `y`, compute the slope `m`, intercept `b`, and use them to predict on a provided test input.
+
+You should implement the closed-form OLS solution:
+
+$$
+m = \frac{\sum_i (x_i - \bar{x})(y_i - \bar{y})}{\sum_i (x_i - \bar{x})^2},\quad
+b = \bar{y} - m\,\bar{x}.
+$$
+
+Then, given `X_test`, output predictions `y_pred = m * X_test + b`.
+
+Return values: `m`, `b`, and `y_pred`.
+

questions/186_linear_regression_ordinary_least_squares/example.json

@@ -0,0 +1,7 @@
+{
+  "input": "X_train = [1, 2, 3]; y_train = [2, 2.5, 3.5]; X_test = [4]",
+  "output": "m = 0.75, b = 1.166667, y_pred = [4.166667]",
+  "reasoning": "Using OLS: m = Cov(X,Y)/Var(X) = 1.5/2 = 0.75 and b = y_bar - m*x_bar = (8/3) - 0.75*2 = 1.166667. Prediction for X_test=[4] is 0.75*4 + 1.166667 = 4.166667."
+}
+
+

questions/186_linear_regression_ordinary_least_squares/learn.md

@@ -0,0 +1,39 @@
+## Learning: Ordinary Least Squares for Simple Linear Regression
+
+### Idea and formula
+- **Goal**: Fit a line $y = m x + b$ that minimizes the sum of squared errors.
+- **Closed-form OLS solution** for 1D features:
+
+$$
+m = \frac{\sum_i (x_i - \bar{x})(y_i - \bar{y})}{\sum_i (x_i - \bar{x})^2},\quad
+b = \bar{y} - m\,\bar{x}
+$$
+
+### Intuition
+- The numerator is the sample covariance between $x$ and $y$; the denominator is the sample variance of $x$.
+- So $m = \operatorname{Cov}(x,y) / \operatorname{Var}(x)$ measures how much $y$ changes per unit change in $x$.
+- The intercept $b$ anchors the best-fit line so it passes through the mean point $(\bar{x},\bar{y})$.
+
+### Algorithm steps
+1. Compute $\bar{x}$ and $\bar{y}$.
+2. Accumulate numerator $\sum_i (x_i-\bar{x})(y_i-\bar{y})$ and denominator $\sum_i (x_i-\bar{x})^2$.
+3. Compute $m = \text{numerator}/\text{denominator}$ (guard against zero denominator).
+4. Compute $b = \bar{y} - m\,\bar{x}$.
+5. Predict: $\hat{y} = m\,x + b$ for any new $x$.
+
+### Edge cases and tips
+- If all $x_i$ are identical, $\operatorname{Var}(x)=0$ and the slope is undefined. In practice, return $m=0$ and $b=\bar{y}$ or raise an error.
+- Centering data helps numerical stability but is not required for the closed form.
+- Outliers can strongly influence OLS; consider robust alternatives if needed.
+
+### Worked example
+Given $X = [1,2,3]$ and $y = [2,2.5,3.5]$:
+
+- $\bar{x} = 2$, $\bar{y} = 8/3$.
+- $\sum (x_i-\bar{x})(y_i-\bar{y}) = (1-2)(2-8/3) + (2-2)(2.5-8/3) + (3-2)(3.5-8/3) = 1.5$
+- $\sum (x_i-\bar{x})^2 = (1-2)^2 + (2-2)^2 + (3-2)^2 = 2$
+- $m = 1.5/2 = 0.75$
+- $b = \bar{y} - m\,\bar{x} = 8/3 - 0.75\cdot 2 = 1.166666\ldots$
+
+Prediction for $X_{test} = [4]$: $y_{pred} = 0.75\cdot 4 + 1.1666\ldots = 4.1666\ldots$
+

questions/186_linear_regression_ordinary_least_squares/meta.json

@@ -0,0 +1,16 @@
+{
+  "id": "186",
+  "title": "Linear Regression via Ordinary Least Squares (OLS)",
+  "difficulty": "hard",
+  "category": "Machine Learning",
+  "video": "",
+  "likes": "0",
+  "dislikes": "0",
+  "contributor": [
+    {
+      "profile_link": "https://github.com/Jeet009",
+      "name": "Jeet Mukherjee"
+    }
+  ]
+}
+

questions/186_linear_regression_ordinary_least_squares/solution.py

@@ -0,0 +1,23 @@
+from typing import List, Tuple
+
+
+def fit_and_predict(X_train: List[float], y_train: List[float], X_test: List[float]) -> Tuple[float, float, List[float]]:
+    n = len(X_train)
+    x_mean = sum(X_train) / n
+    y_mean = sum(y_train) / n
+
+    num = 0.0
+    den = 0.0
+    for i in range(n):
+        dx = X_train[i] - x_mean
+        dy = y_train[i] - y_mean
+        num += dx * dy
+        den += dx * dx
+
+    m = num / den if den != 0 else 0.0
+    b = y_mean - m * x_mean
+
+    y_pred = [m * x + b for x in X_test]
+    return m, b, y_pred
+
+

questions/186_linear_regression_ordinary_least_squares/starter_code.py

@@ -0,0 +1,14 @@
+from typing import List, Tuple
+
+
+def fit_and_predict(X_train: List[float], y_train: List[float], X_test: List[float]) -> Tuple[float, float, List[float]]:
+    """
+    Implement simple linear regression (OLS) to compute slope m, intercept b,
+    and predictions on X_test.
+
+    Returns (m, b, y_pred).
+    """
+    # Your code here
+    pass
+
+

questions/186_linear_regression_ordinary_least_squares/tests.json

@@ -0,0 +1,28 @@
+[
+  {
+    "test": "from questions.186_linear_regression_ordinary_least_squares.solution import fit_and_predict; m,b,y=fit_and_predict([1,2,3],[2,2.5,3.5],[4]); print(round(m,6), round(b,6), [round(v,6) for v in y])",
+    "expected_output": "0.75 1.166667 [4.166667]"
+  },
+  {
+    "test": "from questions.186_linear_regression_ordinary_least_squares.solution import fit_and_predict; m,b,y=fit_and_predict([0,1,2,3],[1,3,5,7],[4,5]); print(round(m,6), round(b,6), [round(v,6) for v in y])",
+    "expected_output": "2 1 [9, 11]"
+  },
+  {
+    "test": "from questions.186_linear_regression_ordinary_least_squares.solution import fit_and_predict; m,b,y=fit_and_predict([0,1,2],[5,2,-1],[3]); print(round(m,6), round(b,6), [round(v,6) for v in y])",
+    "expected_output": "-3 5 [-4]"
+  },
+  {
+    "test": "from questions.186_linear_regression_ordinary_least_squares.solution import fit_and_predict; m,b,y=fit_and_predict([2,2,2],[1,4,7],[10]); print(round(m,6), round(b,6), [round(v,6) for v in y])",
+    "expected_output": "0.0 4.0 [4.0]"
+  },
+  {
+    "test": "from questions.186_linear_regression_ordinary_least_squares.solution import fit_and_predict; m,b,y=fit_and_predict([1,2,3,4],[1.1,1.9,3.05,3.9],[5]); print(round(m,6), round(b,6), [round(v,6) for v in y])",
+    "expected_output": "0.955 0.1 [4.875]"
+  },
+  {
+    "test": "from questions.186_linear_regression_ordinary_least_squares.solution import fit_and_predict; m,b,y=fit_and_predict([3],[7],[10]); print(round(m,6), round(b,6), [round(v,6) for v in y])",
+    "expected_output": "0.0 7.0 [7.0]"
+  }
+]
+
+