10 basic data analytics questions with explanations

1. Which tools do you prefer for data cleaning and preprocessing? Can you describe your process using them?

(This checks familiarity with tools like Python, R, Alteryx, or Tableau Prep for data cleaning. Look for their understanding of
preprocessing steps such as handling missing values, scaling, and encoding.)
Assessment: Listen for specific tools and the process they follow. Good answers will show practical experience.
Expected Answer: "I use Python with Pandas and NumPy for data cleaning. First, I handle missing values by either imputing
them with the mean or dropping rows if too many values are missing. I also standardize data using MinMaxScaler and encode
categorical features with one-hot encoding."

2. What are the different ways you can handle missing data in a dataset? Could you walk me through the steps you’d take?

(This assesses knowledge of different techniques to handle missing data, such as imputation or deletion.)
Assessment: Look for a clear, step-by-step process. They should mention imputation methods or deletion methods.
Expected Answer: "I first check the proportion of missing data. If it's a small amount, I might drop those rows. For numerical
columns, I might impute the missing values using the mean, median, or a predictive model. For categorical columns, I impute
using the mode or use a placeholder like 'Unknown'."

3. How would you perform exploratory data analysis (EDA) using Python or R? Which libraries or frameworks do you use for
this?

(This checks familiarity with EDA libraries and the steps taken to explore data, such as summarizing statistics, visualizing
distributions, etc.)
Assessment: Look for libraries like Pandas, Matplotlib, Seaborn, and tools for statistical summaries and visualizations.
Expected Answer: "I start with Pandas to load and summarize the data. I check for missing values, outliers, and correlations
between variables. Then, I use Seaborn or Matplotlib to create histograms, boxplots, and pair plots to visualize the data
distribution and relationships."

4. What is the syntax to merge two dataframes in Python or R? Can you show an example with both inner and outer joins?

(This tests their ability to merge datasets, which is a key data manipulation skill.)
Assessment: They should demonstrate knowledge of merge() in Python or dplyr::join() in R and should know different types of
joins.
Expected Answer: "In Python, I would use df1.merge(df2, how='inner', on='key') for an inner join and df1.merge(df2,
how='outer', on='key') for an outer join. In R, I use dplyr::inner_join(df1, df2, by='key') for an inner join and dplyr::full_join(df1,
df2, by='key') for an outer join."

5. How do you handle categorical variables in machine learning models? Can you explain the process and syntax in your
preferred tool?

(This assesses understanding of feature encoding techniques like one-hot encoding, label encoding, etc.)
Assessment: Look for knowledge of encoding methods and whether they use libraries like Scikit-learn or R's caret for this task.
Expected Answer: "I typically use one-hot encoding for nominal categorical variables, which can be done using
pd.get_dummies() in Python. For ordinal variables, I might use label encoding with LabelEncoder() from Scikit-learn. If there are
too many categories, I might use target encoding."

6. Can you explain the difference between supervised and unsupervised learning algorithms? Can you give an example of a
project where you used each?

(This tests fundamental machine learning concepts.)

Assessment: Look for clarity in the explanation of the differences and examples of using both types of algorithms.
Expected Answer: "Supervised learning involves labeled data, where the model learns to predict an output variable. For
example, I used linear regression for predicting house prices. Unsupervised learning involves finding hidden patterns in data
without labels, such as clustering customers using k-means."

7. What is your approach to feature selection, and which methods do you prefer when working with large datasets?

(This tests the candidate’s knowledge of feature selection techniques and how to deal with large datasets.)
Assessment: Listen for answers that mention techniques like Recursive Feature Elimination (RFE), feature importance, or
dimensionality reduction (e.g., PCA).
Expected Answer: "I start with domain knowledge and remove features with high correlation. I also use techniques like RFE for
feature selection and PCA for dimensionality reduction, especially when working with large datasets. I also check feature
importance from tree-based models like Random Forest."

8. What is the difference between apply() and map() in Python? Can you provide an example where each would be useful?

(This assesses knowledge of common data manipulation functions in Pandas.)

Assessment: They should explain the difference in how apply() works on DataFrames and map() on Series.
Expected Answer: "apply() is used for applying a function along an axis of a DataFrame (row or column). map() is used for
element-wise transformations on a Pandas Series. For example, I would use apply() to calculate row-wise statistics and map() to
replace values in a Series with specific values."

9. Can you walk me through the process of building a predictive model using Python or R, from data preparation to
evaluation?

(This tests their end-to-end understanding of the machine learning workflow.)

Assessment: They should describe a clear flow from data cleaning to model selection, training, and evaluation.
Expected Answer: "First, I preprocess the data by handling missing values and encoding categorical variables. Then, I split the
data into training and test sets. I select an algorithm (e.g., Random Forest) and train the model using the training set. Afterward,
I evaluate the model using metrics like accuracy, precision, and recall on the test set."

10. How do you visualize the results of your analysis? Which libraries or frameworks do you use to create visualizations in
Python or R?

(This checks for the candidate's ability to use visual tools for presenting data insights.)
Assessment: They should mention visualization libraries and describe the types of visualizations they use.
Expected Answer: "I use Matplotlib and Seaborn for basic visualizations like line plots, histograms, and boxplots in Python. For
more interactive plots, I use Plotly. In R, I prefer ggplot2 for its flexibility and clarity in creating complex visualizations."

Assessment Strategy:

• Technical Understanding: Pay attention to whether the candidate explains tools, methods, and concepts correctly.
Look for technical depth rather than just surface-level knowledge.

• Real-World Application: Ask for examples of projects or scenarios where they've applied these methods. Practical
experience is key.

• Communication: Look for clarity and structure in their answers, especially when explaining complex processes.

By evaluating these aspects, you'll gain insights into the candidate's hands-on experience and their ability to work with tools,
frameworks, and scripts effectively.