Why this exercise?

At Juspay, product managers take end to end product ownership, juggling

between building the product, managing operations and evangelizing our
creation with our customers.

Being a Creator

● Can PMs become TRUE creators?

● Can they CODE? Can they Design?
● Can they analyze data programmatically?

Yes, they do it here! We offer unique opportunities for PMs to upskill

themselves and overlap with Devs and Designers. We believe PMs should
participate in the creation process and make product specifications that
are close to implementation.

While the excel sheet is a tool of choice for many to crunch numbers, it
has its own limitations. A programming tool can empower you to do what you
generally do much more powerfully. Don't worry even if you are programming
for the first time. This document has links to useful resources and some
reference `code` to solve the exercises.

What happened?
On, Feb 14, 2020, you get calls from many merchants. Merchants tell you
that they are seeing a spike in customer complaints and want you to check
if there are any issues. You tell them that you are going to revert to them
in a few hours.

Learning Resources
To identify what exactly happened to these merchants and which other
merchants got impacted and understand the magnitude of the impact, you need
to be able to analyze the data programmatically. Below are some of the
resources that will help you in doing that

R Packages to learn
● ggplot2 (ggplot, geom_*, facet etc )
● dplyr (group_by, summarise, filter etc)
● Optionally R Markdown basics
 Data
You have the payment transactions data for Feb 12, 13 and 14 in a file as
below

Below is a description of different columns in the data set

hr : Start hour of the transaction

mid : Merchant Identifier
t : Total number of transactions
success : Total number of `successful` transactions
pg : Payment Gateway, who collects money on behalf of the merchant
pmt : Payment Method Type such as cards, UPI etc.,
sub_type : The subtype of payment option chosen by the customer
bank : The customer bank account from which the money is sent

Exercise 1

1.1.1 Which dimension combination caused the issue?

Explore the data and visualization to understand when the issue (possibly a
significant number of failures in transactions ) happened and `which
combination of dimension` (pmt, pg, bank and sub_type) has the impact.

Tip: Identify the method to detect an anomaly in a metric across 4+ dimensions

and apply that method to find the above.

1.1.2 When did the issue happen?

What is the `starting hour` and `ending hour` of the reported issue?
Illustrate your answer with a plot

1.1.3 For whom did the issue happen?

Which all `merchants` were impacted because of the issue and which all
merchants were not? Illustrate your answer with a plot and a table.
The following set of questions are focussed on the solutions of the root
cause you have identified. We want you to creatively think and come up with
close to implementation ideas/solutions

1.2 How could we have detected this issue before the merchant
reported? Almost all the merchants expect us to detect the issue before
they detect or their customers detect. What are your thoughts on how this
can be accomplished? Explain with code/pseudo-code.

1.3 How to fix the issue?

Having identified the issue, what are your thoughts on how to fix it so
that there are less customer complaints for the merchants.

Exercise 2
2.1 How to prevent such issues?
The best thing to happen is to prevent the occurrence of these issues. What
are your thoughts on preventing this category of issues?