← BCG Gamma Interview Insights

BCG Gamma·Data Scientist·Technical Phone Screen·Intermediate

IntermediatePrefer not to say
Apr 2026

Summary

BCG Gamma data science case interview built around a transaction dataset. Three questions, all practical ML stuff, nothing too abstract but the second one definitely made me think harder than I expected.

Questions Asked (3)

Q1

Given a transaction dataset with fields like transaction date, client ID, item color, and purchase amount, what features would you engineer to predict whether a client will make a purchase in the next 3 months?

Data ModelingProduct Analytics & Metrics
Author's notes

This is where I spent most of my energy.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the business objective and defining the prediction window and observation window. Then, systematically engineer features from the transaction data across recency, frequency, monetary, and product preference dimensions, ensuring they are computed only from data available before the prediction period to avoid leakage.

Pro tip: Always discuss how you would handle the temporal aspect: use time-based splitting for validation and ensure features are point-in-time correct. This shows you understand real-world deployment challenges.

1. Clarify the problem and define windows

Confirm the prediction horizon (next 3 months) and define the observation window (e.g., past 12 months) for feature engineering. Establish the target variable: whether a client makes at least one purchase in the next 3 months.

2. Engineer RFM features

Compute recency (days since last purchase), frequency (number of transactions), and monetary (total/average spend) for each client. Consider multiple time windows (e.g., last 30, 90, 180 days) to capture different behavioral patterns.

3. Create product preference features

Derive features from item color, such as the proportion of purchases in each color category, color diversity, and whether the client has a dominant color preference. These can indicate taste and potential future purchases.

4. Incorporate temporal and trend features

Calculate inter-purchase time, purchase frequency trends (e.g., increasing/decreasing), and seasonality indicators (e.g., month of last purchase). These help capture customer lifecycle and timing effects.

5. Validate and avoid leakage

Use time-based splitting to validate the model and ensure all features are computed using only data prior to the prediction period. Discuss potential leakage and how to prevent it.

Key Points to Mention

  • Recency, Frequency, Monetary (RFM) features with multiple time windows
  • Product preference features from item color (e.g., color share, diversity)
  • Temporal features: inter-purchase time, purchase trend, seasonality
  • Time-based validation and point-in-time correctness to avoid data leakage
  • Handling of new customers with limited history (e.g., cold-start problem)
  • Business relevance: aligning features with marketing actions (e.g., targeting frequent buyers with color preferences)

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

You train a model and the accuracy looks suspiciously high, almost too good to be realistic. What might have gone wrong?

Root Cause AnalysisProduct Analytics & Metrics
Author's notes

Data leakage.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging that suspiciously high accuracy often signals data leakage or evaluation flaws. Then systematically walk through potential causes, covering data issues, methodology errors, and model validation problems. Conclude by emphasizing the importance of rigorous validation and cross-checking with business logic.

Pro tip: Always validate model performance against a simple baseline and inspect feature importance for leakage indicators. If a feature is too predictive, question its availability at prediction time.

1. Check for Data Leakage

Examine whether any features inadvertently contain information about the target that would not be available at prediction time. Common sources include target encoding, time-based leakage, or duplicated rows.

2. Review Train-Test Split and Validation Strategy

Ensure the data was split correctly (e.g., no overlap, proper stratification, time-based split if needed). Verify that cross-validation folds are independent and that preprocessing steps are fit only on training data.

3. Inspect Evaluation Metrics and Implementation

Double-check that the accuracy metric is computed correctly and appropriate for the problem (e.g., not misleading for imbalanced classes). Look for bugs in code, such as incorrect label mapping or data shuffling errors.

4. Analyze Data Quality and Preprocessing

Look for issues like duplicate records, mislabeled data, or features that are proxies for the target. Check if preprocessing (e.g., scaling, imputation) was applied consistently and without leakage.

5. Compare with Baselines and Sanity Checks

Compare model performance to simple baselines (e.g., majority class) and domain expectations. If the model vastly outperforms, investigate further; also consider if the problem is too easy or the dataset is too small.

Key Points to Mention

  • Data leakage (target leakage, train-test contamination)
  • Improper cross-validation (e.g., not grouping by entity, time-based leakage)
  • Evaluation metric pitfalls (e.g., accuracy misleading for imbalanced data)
  • Data quality issues (duplicates, mislabeled data, outliers)
  • Overfitting to a small or unrepresentative dataset
  • Preprocessing errors (e.g., scaling before split, imputation using full data)

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

Your SVM classifier isn't performing well. What steps would you take to improve it?

Technical Trade-offsAlgorithms & Data Structures
Author's notes

Ran through the usual: kernel choice, regularization parameter C, feature scaling since SVMs are sensitive to that.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by diagnosing the root cause of poor SVM performance through error analysis and validation curves, then systematically address data quality, feature representation, and hyperparameter tuning. Prioritize high-impact fixes like kernel selection and class imbalance before moving to ensemble methods or alternative algorithms.

Pro tip: Always establish a baseline with a simple model (e.g., logistic regression) to confirm whether the issue is SVM-specific or data-related. At BCG Gamma, interviewers value a structured, hypothesis-driven approach over jumping to complex solutions.

1. Diagnose the Problem

Perform error analysis: examine confusion matrix, precision/recall, and misclassified examples to identify patterns (e.g., class imbalance, outliers, non-linear boundaries). Plot learning curves to check for high bias or high variance.

2. Improve Data Quality and Features

Clean data (handle missing values, outliers), engineer new features, and scale features (critical for SVM). Address class imbalance with techniques like SMOTE, class weights, or resampling.

3. Tune SVM Hyperparameters

Optimize C (regularization), kernel type (linear, RBF, polynomial), and kernel-specific parameters (gamma, degree) using grid search or Bayesian optimization with cross-validation.

4. Consider Advanced Techniques

If performance still lags, try ensemble methods (e.g., bagging SVMs), alternative algorithms (e.g., gradient boosting, neural networks), or dimensionality reduction (PCA, feature selection).

5. Validate and Iterate

Use nested cross-validation to avoid overfitting during hyperparameter tuning. Compare against baseline models and iterate based on business metrics (e.g., ROI, accuracy thresholds).

Key Points to Mention

  • Feature scaling (standardization) is crucial for SVM performance
  • Kernel selection: linear for high-dimensional data, RBF for non-linear boundaries
  • Hyperparameters: C (regularization) and gamma (RBF kernel width)
  • Class imbalance handling: class weights, SMOTE, or resampling
  • Cross-validation for reliable performance estimation and hyperparameter tuning
  • Error analysis to guide improvements rather than blind tuning

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.