Data Science Cheat Sheet 2026

The 30 highest-yield Data Science facts, distilled from real exam questions. Print it, save it as a PDF, or study it here — free, no sign-up.

1. Which of the following is a key characteristic of embedded feature selection methods? → The feature selection process is an intrinsic part of the model training algorithm itself.
2. What does the kernel trick enable Support Vector Machines to do? → Map data into a higher-dimensional space to find nonlinear decision boundaries
3. Why scale features before training a k-nearest-neighbors model? → Distance calculations are sensitive to feature magnitude
4. Which of the subsequent functions is utilized for data frame casting? → Dcast
5. A regression model has a low training error but a much higher test error. This is a sign of: → Overfitting
6. Which loss function is most appropriate for a multi-class classification problem with a softmax output layer? → Categorical Cross-Entropy
7. For a multi-class classification problem, macro-averaged F1 differs from micro-averaged F1 because macro: → Treats all classes equally regardless of size
8. Which of the subsequent functions is utilized for flat file loading? → read.table
9. Why standardize features before applying PCA? → So features with larger scales don't dominate the principal components
10. In Bayesian inference, what role does the prior distribution play? → It encodes beliefs about a parameter before observing data
11. For a strongly right-skewed income distribution, which measure of central tendency best represents a typical value? → Median
12. Median imputation is often preferred over mean imputation when the column: → Is skewed or has outliers
13. The Partial Autocorrelation Function (PACF) is primarily used to determine: → The order of the Autoregressive (AR) component
14. A chi-square test of independence is used to assess: → Association between two categorical variables
15. One of the fundamental skills in data science is which of the following? → Machine learning
16. What does the R-squared value indicate in a regression model? → The proportion of variance in the dependent variable explained by the model
17. What does recall (sensitivity) measure? → Of all actual positives, how many were correctly identified
18. What happens when the number of estimators in a Random Forest is increased significantly? → Variance decreases while bias remains roughly constant, with diminishing returns
19. Why might one-hot encoding be necessary before applying linear models to categorical data? → Because models treat numeric category codes as having ordinal magnitude
20. What is a confusion matrix used for? → Evaluating classification model performance
21. Which scenario most likely indicates overfitting? → High training accuracy but low test accuracy
22. Which sampling method gives every member of a population an equal chance of selection? → Simple random sampling
23. Which visualization library is most associated with statistical plotting in Python and is built on Matplotlib? → Seaborn
24. What does the term 'data leakage' refer to in the context of data preparation? → Information from outside the training set improperly influencing model building
25. What is the primary purpose of t-SNE in unsupervised learning? → Visualizing high-dimensional data in 2D or 3D space
26. When building a dashboard for stakeholders, what is the recommended practice for the number of key metrics displayed? → Limit to 5-7 key metrics to avoid cognitive overload
27. Which supervised learning method models the probability of a binary outcome using a logistic (sigmoid) function? → Logistic Regression
28. When should you use a stacked area chart instead of multiple line charts? → When emphasizing cumulative totals and part-to-whole over time
29. Which of the subsequent tests is focused on using data to make decisions? → Hypothesis
30. Binning a continuous variable into discrete intervals is primarily used to: → Reduce the effect of minor observation errors and capture non-linearity