Introduction

There are two tasks, A and B, each one assessing a different skill.

Choose the one you wish to do and send us the results.

TASK A (Churn Classification)

At its core, churn prediction is a classification problem, where the classes often are ‘churned’ and ‘active’. The prediction is based on historical data, including customer behavior, demographics, transaction history, and more.

Goal: well-balanced classification model.

Instructions:

• Work in Python or R.
• Examine the data.
• While working through the layers of challenge, please leave comments in your code.
• Share your solution in a notebook format with us.

Deliverables:

• .ipynb or .R

Rules:

We understand your time is precious and would not want you to spend more than 6 to 8 hours on this over the span of one week max.

Evaluation:

Please send an email with your repository solution. It will be assessed based on:

• Your applied statistics knowledge
• Python or R proficiency
• Your SaaS domain knowledge

TASK B (Sales Visualization)

A large wholesaler of bicycles has noticed that their total profits have been declining over the last few months. They have approached you to help them analyze what are the drivers of this change. The company’s data department has provided you with the company’s sales data over the last few years. You will need to examine this data on your own and decide which key elements you will focus on in your analysis.

Data: Note: the dataset is an open dataset published by Microsoft. We made a condensed version available in a CSV contained inside sales_data.tar.gz file.

Deliverables:

As an outcome of this test, we expect you to send us two documents:

1. A PDF slide deck (<10 slides) detailing your strategic recommendations for the company's executive committee. It should be self-sufficient, i.e. must be understandable by the reader without any further comments. 2, A jupyter-notebook(.ipynb) or .sql or .r file containing the data preprocessing and analysis code.
2. Visuals presented in .twbx file.

As this is based on an open dataset, please refrain from trying to find pre-made solutions online as we would like you to present your own, not someone else's work.

Rules

We understand your time is precious and would not want you to spend more than 6 to 8 hours on this over the span of one week max. The outcome should be runnable locally on a UNIX-flavored OS (MacOS, Linux).

Evaluation:

Send an email with your solution.

It will be assessed based on:

• Your ability to distill strategic recommendations from a large(-ish) data set.
• Your ability to communicate your insights in a clear and simple manner to a target-audience.
• Your ability with data analysis tools (R, Python, or SQL).
• Your ability with Tableau data visualization.

We will review your solution, we strive to get back to you in 1 week. Sometimes it might take more.