This year BCG X grad online assessment (OA) is noticeably more encrypted, and they've started scanning for candidates recently. I completed this set on November 22nd, with four mathematical analysis questions and a total time limit of 90 minutes. Having done over ten sets before, I was familiar with the overall pace and passed all four questions on my first try. Here's a recap of the complete process, the question text, and key points. My impression of this set is: it's not difficult, but it heavily relies on familiarity. There were quite a few pitfalls, especially in the SQL and data processing sections. If this is your first time doing a BCG Online Case Assessment, you'll clearly feel the time pressure. Conversely, if you grasp the patterns, you'll find you can complete all four questions in 30 minutes.
Overall experience: fast-paced, but solid if you're familiar with the routine
There are four separate data manipulation + ML questions, all in the usual BCG X style of Dataset Integration / Feature Processing / Binary Classification. There's not a lot of beating around the bush, it's all about whether or not you're familiar with the full data processing pipeline.
This time, I did it in 30 minutes, including checking, and handed in the paper in the remaining 60 minutes.
If you're doing it for the first time, I'd recommend setting aside at least 60-70min.
Four Questions
Q1 - Multi-source Data Aggregation & Metric Computation
After merging multiple data sources, compute the following metrics.
- Mean value of numerical scores.
- Proportion of boolean language-capability flags.
- Success rate of the order status field.
Return a clean table with the structure: (metric_name, metric_value), ensuring precision and readability.
Q2 - Multi-table Processing & Join Operations
Process three separate datasets.
- Compute driver experience (years).
- Calculate vehicle inspection days.
- Extract trip like counts.
Join the datasets using the primary keys, handle missing values, and select only required fields.
Q3 - Feature Engineering for Numerical & Categorical Variables
Apply feature processing on mixed data types.
- Impute age values using the training-set mean.
- Transform categorical variables into ordinal codes.
- Convert driver-tier categories into binary labels.
Q4 - Model Training with Random Forest Classifier
Merge training and validation sets for full utilization.
Train a Random Forest Classifier to perform binary classification.
Produce the final predicted labels as the output.
Difficulty of the four questions: not difficult, but "stability of details" is examined.
If you are usually familiar with pandas / sklearn, this set is really not that hard. But if you do a lot of BCG X questions, you'll find that they are taken very seriously:
- Be consistent with field naming
- Type conversion cannot be wrong (bool, category, string, numeric)
- The join key should be clean.
- The output format must be strictly adhered to.
That's why it's so easy for newbies to roll over. Especially in the random forest question, if you just fit without merging the training set, you'll get hung up.
My schedule this time.
Just for your information (it's really "proficiency crushing"):
- Q1: 5 minutes
- Q2: 10 minutes
- Q3: 5 minutes
- Q4: 8 minutes
- The rest is all checks.
The whole 30 minutes in one go.
First time in a big factory OA, unfamiliar with the question types?
If you are recently preparing for OA on platforms such as BigFactory Written Exam, HackerRank, CodeSignal, Cowboys.com, etc., we provide Professional level technical assistance services covering the full range of topics such as programming, data processing, SQL, statistical modeling, multi-round numerical scoring questions, etc.
We provide technical assistance remotely online (with secure connection tools such as ToDesk), with no changes or traces to your local environment, to help you maintain a steady pace in high-pressure timing scenarios and avoid missing key steps.
Our purpose is simple:
Charge only if all test cases 100% pass.
If the platform returns results that are not all approved, no fees will be charged. If you need to know the specific assistance or question adaptation, you cancontact us.
