From acdc785d068fc139b7e19575fddafc04f082be00 Mon Sep 17 00:00:00 2001 From: Harshit Saxena <149141289+HARSHIT-SAXENA3007@users.noreply.github.com> Date: Fri, 31 Oct 2025 17:26:02 +0530 Subject: [PATCH 1/3] Added more machine learning quiz questions and explanations Added multiple questions and explanations related to machine learning concepts, including overfitting, gradient descent, dimensionality reduction, evaluation metrics, reinforcement learning, confusion matrix, and neural network challenges. --- machine-learning/machine-learning-quiz.md | 96 +++++++++++++++++++++++ 1 file changed, 96 insertions(+) diff --git a/machine-learning/machine-learning-quiz.md b/machine-learning/machine-learning-quiz.md index 12050acc2d..655174997c 100755 --- a/machine-learning/machine-learning-quiz.md +++ b/machine-learning/machine-learning-quiz.md @@ -1013,3 +1013,99 @@ d) To reduce the complexity of the tree The correct answer is d) To reduce the complexity of the tree. Pruning is a technique used in decision tree learning to simplify the tree by removing branches that do not significantly improve the performance of the model. This is done to prevent overfitting and improve the generalization of the model to new, unseen data. By reducing the complexity of the tree, pruning can help improve the interpretability and efficiency of the model. + +#### Q129. You built a classification model and noticed that it performs very well on the training data but poorly on unseen test data. What problem does your model likely have? + +a) Overfitting +b) Underfitting +c) High bias +d) Low variance + +**Explanation:** +The correct answer is a) Overfitting + +High training accuracy but low testing accuracy indicates that the model memorized rather than generalized — a clear sign of overfitting. + +--- + +#### Q130. When using gradient descent, which of the following will happen if the learning rate is too high? + +a) The algorithm may overshoot the minimum and fail to converge +b) The algorithm will converge more slowly but more accurately +c) The model will underfit the data +d) The gradient will become zero + +**Explanation:** +The correct answer is a) The algorithm may overshoot the minimum and fail to converge + +A large learning rate makes steps too big, causing oscillations or divergence. + +--- + +#### Q131. What is the main goal of dimensionality reduction? + +a) Reduce the number of input features while retaining important information +b) Increase model complexity +c) Improve bias at the cost of variance +d) Remove all correlations between features + +**Explanation:** +The correct answer is a) Reduce the number of input features while retaining important information + +Techniques like PCA aim to compress data by keeping maximum variance with fewer features. + +--- + +#### Q132. Which of these metrics is **not** suitable for evaluating a regression model? + +a) Accuracy +b) RMSE +c) MAE +d) R² + +**Explanation:** +The correct answer is a) Accuracy. + +Accuracy is only meaningful for classification, not regression. + +--- + +#### Q133. You are developing a reinforcement learning model. What is the main component that guides the learning process? + +a) Loss function +b) Reward function +c) Activation function +d) Clustering function + +**Explanation:** +The correct answer is b) Reward function. + +The reward function provides feedback based on actions taken, guiding the agent’s learning. + +--- + +#### Q134. What is a confusion matrix used for? + +a) Evaluating classification model performance +b) Measuring model bias +c) Checking data imbalance +d) Computing correlation + +**Explanation:** +The correct answer is a) Evaluating classification model performance + +It summarizes true/false positives and negatives, helping analyze accuracy, precision, and recall. + +--- + +#### Q135. You trained a neural network and noticed it performs worse as you add more layers. What might be happening? + +a) Vanishing/exploding gradients +b) Overfitting +c) High bias +d) Batch normalization error + +**Explanation:** +The correct answer is a) Vanishing/exploding gradients + +Deep networks may suffer gradient issues that prevent proper learning in earlier layers. From 6290b0b7746560c88ba7525fcc18b157958325e4 Mon Sep 17 00:00:00 2001 From: Harshit Saxena <149141289+HARSHIT-SAXENA3007@users.noreply.github.com> Date: Fri, 31 Oct 2025 22:16:51 +0530 Subject: [PATCH 2/3] Updated machine-learning/machine-learning-quiz.md necessary changes made Co-authored-by: Evgenii Bazhanov --- machine-learning/machine-learning-quiz.md | 1 - 1 file changed, 1 deletion(-) diff --git a/machine-learning/machine-learning-quiz.md b/machine-learning/machine-learning-quiz.md index 655174997c..5e8b962843 100755 --- a/machine-learning/machine-learning-quiz.md +++ b/machine-learning/machine-learning-quiz.md @@ -1026,7 +1026,6 @@ The correct answer is a) Overfitting High training accuracy but low testing accuracy indicates that the model memorized rather than generalized — a clear sign of overfitting. ---- #### Q130. When using gradient descent, which of the following will happen if the learning rate is too high? From 80c2e832aa85197253a5239ab64ce7014774e163 Mon Sep 17 00:00:00 2001 From: Harshit Saxena <149141289+HARSHIT-SAXENA3007@users.noreply.github.com> Date: Fri, 31 Oct 2025 22:21:54 +0530 Subject: [PATCH 3/3] Add new questions and answers to the quiz removed --- --- machine-learning/machine-learning-quiz.md | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/machine-learning/machine-learning-quiz.md b/machine-learning/machine-learning-quiz.md index 5e8b962843..73ef65279b 100755 --- a/machine-learning/machine-learning-quiz.md +++ b/machine-learning/machine-learning-quiz.md @@ -1039,7 +1039,6 @@ The correct answer is a) The algorithm may overshoot the minimum and fail to con A large learning rate makes steps too big, causing oscillations or divergence. ---- #### Q131. What is the main goal of dimensionality reduction? @@ -1053,7 +1052,7 @@ The correct answer is a) Reduce the number of input features while retaining imp Techniques like PCA aim to compress data by keeping maximum variance with fewer features. ---- + #### Q132. Which of these metrics is **not** suitable for evaluating a regression model? @@ -1067,7 +1066,7 @@ The correct answer is a) Accuracy. Accuracy is only meaningful for classification, not regression. ---- + #### Q133. You are developing a reinforcement learning model. What is the main component that guides the learning process? @@ -1081,7 +1080,7 @@ The correct answer is b) Reward function. The reward function provides feedback based on actions taken, guiding the agent’s learning. ---- + #### Q134. What is a confusion matrix used for? @@ -1095,7 +1094,6 @@ The correct answer is a) Evaluating classification model performance It summarizes true/false positives and negatives, helping analyze accuracy, precision, and recall. ---- #### Q135. You trained a neural network and noticed it performs worse as you add more layers. What might be happening?