50 IBM A1000-041 - Assessment: Data Science Foundations - Level 1 Practice Questions: Question Bank 2025

A retailer wants to reduce customer churn. Stakeholders first ask, “What exactly are we trying to achieve and how will we measure success?” In the Data Science Methodology, which phase does this correspond to?

You are exploring a dataset with two numeric variables: weekly advertising spend and weekly sales. Which visualization is most appropriate to assess whether they have a linear relationship?

In Python, you have a list named values = [3, 7, 2, 7]. Which expression returns 7 (the first occurrence) as an index?

A team builds a model to predict whether a transaction is fraudulent (fraud vs not fraud). Which type of machine learning problem is this?

A data scientist notices many missing values in a key feature and discovers they occur mostly for one specific data source system. What is the BEST next step in the Data Science Methodology?

You are comparing the distribution of customer ages across three customer segments. Which visualization is most effective for comparing distributions across groups?

A pandas DataFrame df contains a column 'income'. Some entries are strings like '50000' and others are numeric. A model training step fails due to mixed types. What is the most appropriate fix?

A binary classifier achieves 98% accuracy, but only 1% of transactions are actually fraud. The model rarely flags fraud. Which metric is most appropriate to evaluate how well fraud cases are being identified?

A data science team repeatedly evaluates multiple models and feature sets on the test dataset, selecting whichever performs best. Later, performance drops significantly in production. What is the most likely cause?

You are designing an end-to-end approach to build a predictive model. You have limited labeled data and suspect that collecting labels is expensive. Which strategy best aligns with sound data science methodology to reduce wasted effort before heavy modeling?

A retail team wants to reduce returns by understanding the most common reasons customers return products. They have thousands of free-text return comments. What is the most appropriate first step in a Data Science Methodology approach?

You are using pandas to select a subset of rows where the column "age" is greater than 30 and the column "country" equals "CA". Which option correctly performs this filtering?

A bar chart comparing average sales across 12 product categories shows labels overlapping and becoming unreadable. What is the most effective visualization adjustment?

A data scientist is evaluating a binary classifier for rare fraud events (1% positive class). Accuracy is 99%, but the model misses most fraud cases. Which metric is most appropriate to prioritize for detecting fraud cases?

During data understanding, you find a numeric feature with a long right tail (highly skewed). Many models perform better when the distribution is less skewed. Which transformation is a common, appropriate approach?

A team wants to compare a model’s performance across multiple train/test splits without leaking information from the test sets. Which approach best supports this goal?

A dataset contains missing values in a column "income". You suspect missingness is not random (e.g., higher-income individuals are less likely to report). What is the most appropriate next action before choosing an imputation strategy?

A colleague writes the following code to add a new column to a DataFrame: df['rate'] = df['sales'] / df['visits'] but gets infinite values. Which is the most likely cause and best fix?

You are building a model with one-hot encoded categorical variables and a regularized linear model (e.g., logistic regression). You notice that features with larger numeric ranges dominate optimization. What is the best practice to address this?

A data science team has built a churn model and achieved high AUC on historical data. However, when deployed, performance degrades over time. Which architecture/operational approach best helps detect and respond to this issue?

A retail team has defined the problem, gathered data, and built an initial model. Stakeholders now ask whether the solution will remain reliable as customer behavior changes seasonally. Which Data Science Methodology activity best addresses this concern?

You are exploring a dataset with two numeric variables and want to quickly check whether their relationship is approximately linear and identify potential outliers. Which visualization is most appropriate?

A DataFrame column contains values like "$1,200", "$350", and "N/A". You need to compute summary statistics for this column. What is the best first step in Python?

A binary classifier shows 96% accuracy on a dataset where only 4% of cases are positive. The business cares most about capturing positives, even at the cost of more false alarms. Which metric is the most appropriate to prioritize?

A project team is unsure whether they should proceed with collecting new data because they may not need it. According to a disciplined Data Science Methodology, what should be done first?

A team wants to compare the distribution of a numeric variable (e.g., transaction amount) across several categories (e.g., store regions) and quickly spot differences in medians and outliers. Which plot is best?

A notebook runs without errors, but a function that updates a counter variable always returns 1 instead of increasing. The code looks like: count = 0 def increment(): count = count + 1 return count What is the most likely cause and fix?

You are training a model and observe very low training error but significantly higher validation error. Which situation best explains this pattern?

A bank is building a credit-risk model and must explain individual decisions to auditors. The dataset includes demographic attributes that could introduce bias. Which approach best supports both interpretability and responsible use?

A dashboard shows average delivery time by city. Some cities have only a few deliveries, causing highly volatile averages and misleading conclusions. What is the best practice to address this issue in the analysis/visualization?

A retail team has a project goal: “Reduce customer churn.” They are unsure what data to request and how success will be measured. Which Data Science Methodology step should they focus on first to clarify the objective and analytic approach?

You plot a histogram of transaction amounts and notice a long right tail with a few extremely large values. Which preprocessing step is often appropriate before applying methods sensitive to scale (e.g., distance-based algorithms)?

In pandas, you want to select rows where the column "age" is greater than 30. Which expression correctly filters the DataFrame df?

A data scientist is evaluating a binary classifier for rare fraud cases (1% positive). Accuracy is very high, but many frauds are missed. Which metric is typically more informative for this situation?

During data preparation, you find missing values in a "salary" feature. You plan to train a linear regression model. Which approach is a generally acceptable baseline for handling missing numeric values?

A project team built a model that performs well in testing, but business stakeholders do not trust it because they cannot understand the features driving predictions. Which action best aligns with a foundational best practice to address this concern?

You attempt to merge two pandas DataFrames on customer_id, but the result has far more rows than expected. Which issue most commonly causes this type of row explosion?

You want to validate a classification model and reduce the chance that a single train/test split gives an overly optimistic estimate. Which technique best addresses this?

A model achieves excellent results during development. After deployment, performance degrades because incoming data has different distributions (e.g., a new customer segment). What is the most appropriate concept describing this issue?

A team is building a customer churn model. They include a feature called "churn_date" (the date the customer canceled) and observe unusually high performance. What is the most likely problem, and what is the best corrective action?

A team is defining a data science project to reduce customer churn. Multiple departments suggest potential predictors, but the business sponsor wants clarity on what “success” looks like. What is the best next step in the Data Science Methodology?

You receive a dataset with a categorical column 'priority' containing values: 'Low', 'Medium', 'High'. For a linear model, you want to avoid implying an arbitrary numeric distance between categories. Which encoding is most appropriate?

In a Jupyter notebook, a DataFrame column contains missing values coded as NaN. You want to count how many missing values are in that single column. Which approach is most appropriate?

A retail analyst creates a scatter plot of advertising spend vs. sales, but most points are compressed into a dense blob because a few campaigns have extremely high spend. What is a good first visualization adjustment to reveal structure for typical campaigns?

A dataset has two numeric features measured in different units (e.g., income in dollars and age in years). You plan to use a distance-based algorithm such as k-NN. What preprocessing step is most important?

During EDA, you discover that a “transaction_amount” column is heavily right-skewed with a long tail. You want a transformation that often makes the distribution more symmetric and can stabilize variance. What is a common choice?

A project team has built a model but realizes the training data is not representative of the population where the model will be used (different region and customer mix). According to the data science methodology, what is the best action?

You are analyzing a binary classification dataset with 2% positive cases. Accuracy is very high even for a naive model that predicts all negatives. Which metric is generally more informative in this situation?

A data scientist fits a model and evaluates it on the same dataset used for training, obtaining excellent results. Later, performance drops sharply on new data. What issue most likely occurred?

A notebook prepares features using the full dataset (including the test set) to compute normalization parameters (mean/std), then trains and evaluates a model. The test score looks suspiciously high. What is the best fix?