Master of Data science Exploratory Data Analysis 1 Flashcards
6 cards from real MS-DS Master of Data science practice questions. Tap to flip, then mark Knew It or Still Learning — missed cards come back until you master them.
Read the first 6 Master of Data science Exploratory Data Analysis 1 flashcards as text
Which statistical measure is most resistant to the influence of outliers when describing the center of a dataset?
Answer: Median
The median is resistant to outliers because it depends only on the middle value(s) of an ordered dataset, not the magnitude of extreme values. The mean, by contrast, is pulled toward outliers.
A histogram of a feature shows that most values cluster near zero with a long tail extending to very large positive values. What transformation would most likely normalize this distribution?
Answer: Square root or logarithmic transformation
Log or square root transformations compress large values and spread small values, which is effective for right-skewed distributions common when data clusters near zero with a long positive tail. Standardization and scaling preserve shape.
During EDA, you notice that two features have a Pearson correlation coefficient of –0.87. What does this indicate?
Answer: As one feature increases, the other tends to decrease strongly
A Pearson correlation of –0.87 indicates a strong negative linear relationship: as one variable increases, the other tends to decrease. Correlation does not imply causation, and Pearson only measures linear association.
What is the primary purpose of a box plot in exploratory data analysis?
Answer: To summarize the distribution of a continuous variable and highlight outliers
A box plot visualizes the five-number summary (minimum, Q1, median, Q3, maximum) and flags potential outliers as points beyond 1.5×IQR from the quartiles, making it effective for understanding spread, skew, and anomalies.
In EDA, what does a high proportion of missing values in a feature column most directly suggest to an analyst?
Answer: The feature may be systematically missing and requires investigation before deciding on imputation or removal
High missingness warrants investigation into whether the data is missing at random, missing completely at random, or missing not at random. The mechanism determines whether imputation, removal, or alternative handling is appropriate.
Which plot is best suited for visualizing the distribution of a single continuous variable and detecting potential bimodality?
Answer: KDE (Kernel Density Estimate) plot
A KDE plot smooths the data into a continuous probability density curve, making it easy to see whether the distribution has one peak (unimodal) or two (bimodal). Histograms can also reveal this but are sensitive to bin width; KDE provides a cleaner view.