Some quick data challenges.
You belong to the data team at a local research hospital. You've been tasked with developing a means to help doctors diagnose breast cancer. You've been given data about biopsied breast cells; where it is benign (not harmful) or malignant (cancerous).
- What features of a cell are the largest drivers of malignancy?
- How would a physician use your product?
- There is a non-zero cost in time and money to collect each feature about a given cell. How would you go about determining the most cost-effective method of detecting malignancy?
Yammer’s Analysts are responsible for triaging product and business problems as they come up. In many cases, these problems surface through key metric dashboards that execs and managers check daily.
You show up to work Tuesday morning, September 2, 2014. The head of the Product team walks over to your desk and asks you what you think about the latest activity on the user engagement dashboards.
Start to work your way through your list of hypotheses in order to determine the source of the drop in engagement.
Answer the following questions:
- Do the answers to any of your original hypotheses lead you to further questions?
- If so, what are they and how will you test them?
- If they are questions that you can’t answer using data alone, how would you go about answering them (hypothetically, assuming you actually worked at this company)?
- What seems like the most likely cause of the engagement dip?
- What, if anything, should the company do in response?
Company XYZ sells a software for $39. Since revenue has been flat for some time, the VP of Product has decided to run a test increasing the price. She hopes that this would increase revenue. In the experiment, 66% of the users have seen the old price ($39), while a random sample of 33% users a higher price ($59).
The test has been running for some time and the VP of Product is interested in understanding how it went and whether it would make sense to increase the price for all the users.
Especially he asked you the following questions:
- Should the company sell its software for $39 or $59?
- The VP of Product is interested in having a holistic view into user behavior, especially focusing on actionable insights that might increase conversion rate. What are your main findings looking at the data?
- [Bonus] The VP of Product feels that the test has been running for too long and he should have been able to get statistically significant results in a shorter time. Do you agree with her intuition? After how many days you would have stopped the test? Please, explain why.
Time limit: 4hrs.
Goals:
You are a data scientist at a large construction company. In order to submit more competitive bids for contracts, the head of analytics and data science would like you to work on forecasting core construction material costs. You have access to the pricing information (per unit, in dollars) for the company’s past purchases of plywood, sheetrock, steel beams, rebar, and glass contracts over the past ten years in csv format .
Currently, the company is able to purchase futures contracts in a material for up to six months in advance. Thus, your manager is most interested in forecasting the price of each commodity up to six months into the future. In particular, your manager feels it would be really useful to understand when a commodity price is about to increase temporarily, in order to either buy the commodity early or wait until after the temporary increase.