第 4 / 5 页
非代码面试题
显示 20 / 82 道匹配题目
答题状态：未尝试未正确已正确
ID题目领域难度题型进度权限
4381More Configurations, More OptimismFold scores vary wildly because different time periods behave very differently. What is the first tuning response you should consider before trusting a single mean CV number?机器学习中等essay未尝试面试订阅4382Search Space WidthThe best score in your search occurs at the largest regularization value on the grid. What does that suggest as the next tuning step?机器学习中等essay未尝试面试订阅4383Noisy Metric, Narrow BudgetAs model capacity increases, training performance keeps improving but validation performance stays flat. From a tuning perspective, what direction should you test next?机器学习中等essay未尝试面试订阅4384Nested Versus Flat EstimateA random search keeps finding similar good values over a broad region of hyperparameters. What does that usually suggest about the marginal value of making the search grid much denser there?机器学习中等essay未尝试面试订阅4385Early-Stopping PatienceOne CV fold is much smaller than the others and dominates the variance of the average score. What tuning-related design concern should you address first?机器学习中等essay未尝试面试订阅4386Before TuningTraining AUC is very high but CV AUC is near chance. Before trying more hyperparameter values, what is the first diagnostic step?机器学习中等essay未尝试面试订阅4387Before Adding More DimensionsTwo hyperparameter settings differ in mean CV score by only 0.001, while the estimated standard error is 0.010. What is the first sensible interpretation?机器学习中等essay未尝试面试订阅4388Before Reporting Best ScoreA time-series tuning run says a very short lookback window wins, but recent live performance has deteriorated sharply. What should you inspect first before widening the search?机器学习中等essay未尝试面试订阅4389Before Choosing Nested CVA categorical encoder was fit once on all rows and then reused inside cross-validation. What is the immediate correction?机器学习中等essay未尝试面试订阅4390Before Blaming The SearchThe current best setting sits at extreme values on both the learning-rate and regularization grids. What should your next search action be?机器学习中等essay未尝试面试订阅5041Recover the Missing Fold Gap 1A 5-fold cross-validation comparison records four paired score differences (model A minus model B): [0.02, 0.01, -0.01, 0.03]. The desk report says the overall mean fold difference across all 5 folds was 0.01. What was the missing fifth-fold difference?机器学习中等数值题未尝试面试订阅5042Recover the Missing Fold Gap 2A 5-fold cross-validation comparison records four paired score differences (model A minus model B): [0.05, 0.02, 0.04, -0.01]. The desk report says the overall mean fold difference across all 5 folds was 0.026. What was the missing fifth-fold difference?机器学习中等数值题未尝试面试订阅5043Recover the Missing Fold Gap 3A 5-fold cross-validation comparison records four paired score differences (model A minus model B): [-0.02, 0.01, 0.0, -0.01]. The desk report says the overall mean fold difference across all 5 folds was 0.002. What was the missing fifth-fold difference?机器学习中等数值题未尝试面试订阅5044Recover the Missing Fold Gap 4A 5-fold cross-validation comparison records four paired score differences (model A minus model B): [0.01, 0.01, 0.02, 0.0]. The desk report says the overall mean fold difference across all 5 folds was 0.014. What was the missing fifth-fold difference?机器学习中等数值题未尝试面试订阅5045Recover the Missing Fold Gap 5A 5-fold cross-validation comparison records four paired score differences (model A minus model B): [0.04, -0.02, 0.01, 0.02]. The desk report says the overall mean fold difference across all 5 folds was 0.01. What was the missing fifth-fold difference?机器学习中等数值题未尝试面试订阅5046Recover Discordant Counts From McNemar Summary 6Two classifiers are compared on the same test set. The total number of discordant cases is b+c=16, model A is better so b>c, and the continuity-corrected McNemar statistic is 3.0625. What discordant counts (b,c) are implied?机器学习中等数值题未尝试面试订阅5051Cost-Sensitive Deployment Choice 11Model A makes 8 false positives and 2 false negatives on a validation set. Model B makes 6 false positives and 5 false negatives. If a false negative costs 10 units and a false positive costs 1 unit(s), which model has lower expected validation cost and what are the two costs?机器学习中等数值题未尝试面试订阅5061Why Nested Evaluation MattersWhy is it unfair to compare two tuned models on the same validation folds that were also used to pick their hyperparameters?机器学习困难essay未尝试面试订阅5062Why Dependence MattersWhy can standard iid significance arguments be too optimistic when model scores come from overlapping rolling windows?机器学习困难essay未尝试面试订阅5063Metric Choice Can Flip WinnerWhy can a model that wins on AUC still lose on business utility?机器学习困难essay未尝试面试订阅