Which visualization is best for showing data distribution?HistogramLine chartPie chartScatter plotA) HistogramHistograms show frequency distribution of continuous data.
Correlation measures:Cause and effectStrength of linear relationshipVariability in dataPrediction accuracyB) Strength of linear relationshipCorrelation quantifies the relationship between two variables.
P-value in hypothesis testing represents:Probability of rejecting null hypothesisProbability of observing results by chanceType II error rateConfidence intervalB) Probability of observing results by chanceP-value shows likelihood of obtaining results assuming null hypothesis is true.
Which machine learning algorithm is based on Bayes’ theorem?Decision TreesNaïve BayesRandom ForestK-meansB) Naïve BayesNaïve Bayes uses Bayes’ theorem for probabilistic classification.
Standard deviation measures:Central tendencySpread of dataSkewnessSample sizeB) Spread of dataStandard deviation shows how much data deviates from the mean.
Which graph is best for showing categorical data distribution?Line chartHistogramBar chartScatter plotC) Bar chartBar charts represent categorical data with rectangular bars for each category.
P-value in hypothesis testing indicates:Sample sizeProbability of rejecting null hypothesisProbability of observing data under null hypothesisConfidence intervalC) Probability of observing data under null hypothesisA small p-value suggests strong evidence against the null hypothesis.
Which measure is resistant to outliers?MeanMedianVarianceStandard deviationB) MedianMedian is not affected by extreme values.
In regression, multicollinearity occurs when:Predictors are highly correlatedResiduals are normally distributedIndependent variables are uncorrelatedErrors are independentA) Predictors are highly correlatedMulticollinearity reduces reliability of regression coefficients.
Which distribution models rare events?NormalPoissonBinomialUniformB) PoissonPoisson distribution is used for modeling rare or infrequent events.
P-value measures:Probability null hypothesis is trueProbability of observed result under nullVariance in dataEffect sizeB) Probability of observed result under nullLow p-value indicates evidence against null hypothesis.
Overfitting occurs when a model:Performs well on training but poorly on test dataPerforms equally on bothHas low varianceIgnores noiseA) Performs well on training but poorly on test dataOverfit models capture noise, harming generalization.
Standard deviation indicates:Central tendencyData spread around meanMedianSkewnessB) Data spread around meanIt quantifies variability of data values.
Which technique reduces dimensionality?RegressionPCAClusteringClassificationB) PCAPrincipal Component Analysis projects data into fewer dimensions.
F1-score balances:Accuracy and recallPrecision and recallVariance and biasMean and varianceB) Precision and recallF1 is harmonic mean of precision and recall.
The mean of dataset {2, 4, 6, 8} is:4568B) 5Mean = sum / count = (20 / 4) = 5.
Which distribution is bell-shaped?Normal distributionUniform distributionPoisson distributionExponential distributionA) Normal distributionNormal distribution is symmetric and bell-shaped.
Correlation coefficient ranges between:–1 and +10 and 1–∞ to +∞–10 to +10A) –1 and +1It measures strength and direction of relationships.
Which metric handles class imbalance?AccuracyF1-scoreMean absolute errorR² scoreB) F1-scoreF1 balances precision and recall.
Outliers are best detected using:BoxplotsHistogramsMeanModeA) BoxplotsBoxplots visually highlight extreme values.
In hypothesis testing, the p-value represents:Probability null hypothesis is trueProbability of observing result under null hypothesisSample mean) Effect sizeB) Probability of observing result under null hypothesisp-value quantifies evidence against null hypothesis.
A confusion matrix is used to evaluate:Regression modelsClassification modelsClustering modelsStatistical distributionsB) Classification modelsIt summarizes TP, FP, TN, FN results.
Which distribution is symmetric and bell-shaped?) Normal distributionPoisson distributionExponential distributionGeometric distributionA) Normal distributionNormal distribution is the most common continuous distribution.
Cross-validation helps by:Increasing dataset sizeEstimating model generalizationReducing noiseScaling featuresB) Estimating model generalizationCV tests model performance on unseen folds.
The process of removing noise and inconsistencies from data is:Data VisualizationData CleaningData ReductionData EncodingB) Data CleaningCleaning ensures high-quality data for analysis.
In statistics, p-value represents:Probability of null hypothesis being trueProbability of observing results by chanceCorrelation strengthRegression coefficientB) Probability of observing results by chanceLower p-values indicate stronger evidence against null hypothesis.
A confusion matrix is used in:RegressionClassification evaluation: ClusteringSamplingB) Classification evaluationIt measures performance via true/false positives and negatives.
Feature scaling ensures:Faster convergence during trainingOverfitting reduction onlyFeature deletionLabel encodingA) Faster convergence during trainingScaling standardizes feature ranges for better performance.
P-value < 0.05 indicates:Strong evidence against nullAccept null hypothesisRandom variationModel overfitA) Strong evidence against nullSuggests statistically significant result.
Z-score measures:Distance from mean in SD unitsMedian shift Sample sizeSkewnessA) Distance from mean in SD unitsZ = (X−μ)/σ standardizes values.
Outliers most affect:MeanMedianMode Variance onlyA) MeanMean shifts heavily due to extreme values.
K-Means algorithm minimizes:Inter-cluster distanceIntra-cluster variance GradientCovarianceB) Intra-cluster varianceClusters points near centroids.
In Data Science, the process of detecting and correcting errors in data is called:Data IntegrationData CleaningData ModelingData ExtractionB) Data CleaningData cleaning removes inconsistencies and ensures data quality before analysis.
The “p-value” in hypothesis testing represents:Probability that null hypothesis is trueProbability of observing results as extreme as the actual ones, assuming null hypothesis is trueStatistical mean of dataConfidence intervalB) Probability of observing results as extreme as actual onesA smaller p-value (<0.05) suggests strong evidence against the null hypothesis.
Which visualization is most effective for showing the relationship between two numerical variables?Bar chartPie chartScatter plotHistogramC) Scatter plotScatter plots reveal correlations or patterns between two quantitative features.
Which metric measures model accuracy for regression problems?R² scoreF1 scorePrecisionRecallA) R² scoreR² represents the proportion of variance explained by the model.
What is feature scaling used for?Normalizing data valuesEncoding categoriesRemoving missing dataDetecting outliersA) Normalizing data valuesIt ensures features contribute equally to model learning.
What is the main goal of hypothesis testing?Predicting outcomesVerifying sample validityDetermining statistical significanceEstimating population meanC) Determining statistical significanceHypothesis testing evaluates evidence to support or reject assumptions.
Which technique reduces dimensionality?PCASVMK-MeansDecision TreeA) PCAPrincipal Component Analysis transforms data into fewer uncorrelated variables.
What is overfitting?Model learns training data too wellData under-samplingMissing variablesNoisy inputs onlyA) Model learns training data too wellOverfitting occurs when the model performs well on training but poorly on new data.
Mean, Median, and Mode are types of:Probability modelsCentral tendencyDispersion measuresSampling techniquesB) Central tendencyThey summarize the center of data distribution.
Which test checks if two groups’ means are significantly different?Chi-squareT-testANOVARegressionB) T-testT-test compares averages between two samples.
Outliers affect which statistic the most?MeanMedianModeCountA) MeanExtreme values distort the arithmetic average.
In machine learning, "feature scaling" is done to:Increase accuracyEqualize variable rangesRemove missing dataSplit datasetsB) Equalize variable rangesScaling ensures fair influence across features.
The process of cleaning and organizing data is called:Data transformationData wranglingData mining Data validationB) Data wranglingData wrangling prepares raw data for analysis.
In hypothesis testing, the p-value indicates:Population sizeProbability of observing result by chanceVarianceBias levelB) Probability of observing result by chanceA small p-value shows strong evidence against the null hypothesis.
The process of finding hidden patterns in data is called:Data cleaningData filteringData mining Data normalizationC) Data miningData mining extracts meaningful insights and trends from data.
Which distribution is symmetrical and bell-shaped?UniformPoissonNormalExponentialC) NormalThe normal distribution is widely used in statistics and ML.
Correlation measures:Dependency between variablesVariance of sampleMean valueClass countA) Dependency between variablesCorrelation quantifies the linear relationship between two variables.
What does the p-value represent in hypothesis testing?Probability of errorProbability of obtaining observed results under the null hypothesisConfidence levelSignificance limitB) Probability of obtaining observed results under the null hypothesisA low p-value indicates strong evidence against the null hypothesis.
Which measure of central tendency is affected most by outliers?MeanMedian ModeRangeA) MeanThe mean changes drastically with extreme values, unlike the median or mode.
What is “feature scaling” used for in machine learning?Normalizing data rangeAdding new variablesEncoding categoriesRemoving duplicatesA) Normalizing data rangeFeature scaling ensures numerical features contribute equally to model training.
What is overfitting in a statistical model?Poor performance on training dataExcellent generalizationModel learns noise instead of patternData imbalanceC) Model learns noise instead of patternOverfitted models perform well on training data but fail on unseen data.
In a normal distribution, mean = median = mode indicates:SkewnessSymmetryKurtosis OutliersB) SymmetryEqual mean, median, and mode signify a perfectly symmetrical bell curve.
What does correlation measure?CausalityStrength of linear relationshipFrequency of dataModel accuracyB) Strength of linear relationshipCorrelation quantifies how strongly two variables move together.
What is the main purpose of hypothesis testing?Generate dataMake predictionsTest assumptions about a populationReduce data sizeC) Test assumptions about a populationIt helps determine if results are statistically significant.
Which measure shows the spread of data around the mean?MeanMedianStandard deviation ModeC) Standard deviationIt quantifies how much data varies from the mean.
What is overfitting in data models?Model fits only training dataModel performs equally on all sets Model ignores data trendsModel predicts random resultsA) Model fits only training dataOverfitted models fail to generalize to unseen data.
Which technique reduces dataset dimensionality?KNNPCARegression Decision treeB) PCAPrincipal Component Analysis compresses data by removing correlated features.
Which visualization is best for categorical comparison?HistogramPie chartScatter plotLine chartB) Pie chartPie charts show the proportion of categories in a dataset.
The mean of data is affected by:All observations Only extreme valuesMedianNoneA) All observationsMean considers all data points, making it sensitive to outliers.
Overfitting in data science occurs when:The model learns noiseData is incompleteDataset is small Features are fewA) The model learns noiseOverfitting means the model performs well on training data but poorly on new data.
Which metric measures model accuracy for classification?MAEMSEPrecisionR²C) PrecisionPrecision shows the ratio of correctly predicted positives to total predicted positives.
Feature scaling helps:Normalize data valuesRemove missing dataIncrease dataset sizeMerge datasetsA) Normalize data valuesScaling ensures that features contribute equally to distance-based models.
Correlation measures:Difference between meansRelationship strength VarianceSkewnessB) Relationship strengthIt quantifies how strongly two variables are linearly related.
In regression analysis, multicollinearity occurs when:Variables are independentPredictors are highly correlatedErrors are random Data is normalizedB) Predictors are highly correlatedMulticollinearity reduces model interpretability and accuracy.
A confusion matrix is used to evaluate:ClusteringClassification Regression ForecastingB) ClassificationIt displays actual vs. predicted outcomes for classification models