rockaway Hackathon 2017

IBM Cloud

We call upon the development teams. Grab your laptops, warm-up your processors and brains. You will get 24 hours.
Winner takes all – exactly 100.000 CZK.

Using the IBM Cloud technology you will solve a complex e-commerce problem. Your solution must be simple, efficient and, above all, functional - so that we can engage in the Mall Group infrastructure. Experienced mentors eager to help you squeeze out the maximum both from your team and technologies will be available non-stop. And the winner takes 100.000 CZK.

Where will we meet?

Kick-off Meeting

November 20th 2017
IBM Prague
V Parku 2294/4
Attention: At least 1 member of the team
must personally participate in Kick-off!
Event

Hackathon Prague

December 1st-2nd 2017
IMPACT HUB Praha
Drtinova 10
By registering, you agree to the terms of the
competition. You can find them below in the FAQ.
Event

Are you in?

Sign up

Any questions?

Go to FAQ

Schedule

All times are specified approximately. The faster we will deal with organizational details, the more time will remain for hacking.

Friday

17:30 – 18:00Registration
18:00 – 18:30What it's all about
18:30 – 18:45Creating teams
18:45 – 19:00Checking that everyone knows what and how to do
19:00 – 22:00Hacking
22:00 - 22:30Late dinner
22:30 – ?Hacking

Saturday

08:00 – 08:30Breakfast
08:30 – 09:00Enlightenment
09:00 – 10:30Hacking
10:30 – 11:00Coffeebreak
11:00 – 13:00Hacking
13:00 – 14:00Lunch
14:00 – 17:00Hacking
17:00 – 17:30Coffeebreak
17:30 – 21:30Hacking
21:30 – 22:00Presentations and announcement of the winners

Assignment

After many reflections, we chose the 4 themes that we found the most appropriate. At first glance, they may look simple, but we are looking for great and perfect solutions that we can use.
It will be fun for programmers, but also for data engineers, mobile application developers and products.ci.

Anomaly detection

What’s going on:
You will receive data from Google Analytics (a tool for tracking user behaviour on the web that will allow you to show, for example, the number of transactions per period for the sessions of users who came to our site from Heureka.cz). Your goal is to develop an application (script / webapp / jupyter notebook / …) that will allow an analyst to automatically detect the specifics (events) in these data.
Since these data have multiple metrics and attributes, it is necessary to report the detected event with respect to a specific data cut (ie a specific combination of metrics, attributes, and granularity).

Here are some types of events that might interest the user:
* extreme values
* Significant changes in trend
* unexpected monotony (e.g., a gradual drop on a cut where volatility is typical)
* Changing the link to another cut

What metrics we are interested in:
* Number of transactions
* Revenue (ie transaction value in CZK)
* Conversion Rate
* Add2cart CR (how many percents of customers will go from product detail view to shopping cart)
* Cart through rate (how many percents of customers will pass the thank-you page basket)
* Number of sessions

Definition of done:
Must have features:
* Most (ideally all) events should be quantifiable primarily in money
* The user should be able to set the area where the anomaly wants to look
* mutual prioritization of detected events – I show the most fundamental issues at the top
* Events are shown at the granularity level of the day / week / month / year

Nice to have features:
* outputs can be exported (csv, save plot)
* The user parameterizes the severity of individual modules (eg pval / std for anomaly detection)

Why implement it? Current issue:
We have a lot of data that we can not walk through and report with standard Tableau, GoodData, and we know that our attention is being missed by a lot of interesting events, the knowledge of which will save us a lot of costs. We want a tool that will search for and distribute part of such events for us.

Benefits for e-shop:
Cost savings thanks to timely intervention for critical events.

Data and inputs:
Data frame from Google Analytics

Supervisor:
Viktor Šohájek

Visual Search

What’s going on:
We want to make it easier for customers to search for products by capturing a product through their mobile phone and then displaying the same or similar products.
For example, the customer will photograph a coffee machine. Visual search from the image correctly recognizes that it is a coffee maker. It also determines that it is a capsule coffee maker, black, Nespresso type. The customer will show a coffee shop page where only Nespresso black capsules are filtered out.
The solution will only be available on tablets and mobile phones.

Why implement it? Current issue:
The customer often fails to correctly name a certain type of goods (such as hoverboards, elliptical trainers, sports testers, …) he wants to find. Alternatively, he can name the type of product, but he needs to enter a large number of product parameters (e.g., Addidas women’s black leather boots, black colour) for a proper search, which is annoying to him.

Benefit for our customer:
Faster and simpler product search and subsequent selection.

Benefits for e-shop:
Increase the volume of orders.

Data and inputs:
Neural network recognizing general entities in the image (IBM solution). Product Database at Mall.cz (XML Product Feed).

Supervisor:
Roman Dušek, Michal Moravec

Tip for a better product

What’s going on:
The customer decides to buy some goods and does not know that if he paid a little more, he could buy better goods. The purpose of this task is to offer the customer an alternative product for the currently being viewed, including a clear declaration of how much the customer has to pay and what he gets. For example, a customer chooses an LCD TV and we warn him that if he pays for 500 CZK more, he gets the same TV, including the built-in DVD player. It is necessary to analyze the attributes and properties of the products, to compare them (some will need to be fixed, for example, if the client chooses 44 size shoes, we will not offer different sizes) and recommend a more profitable/interesting product. Subsequent recommendations for available alternatives may have several variants such as: for the same price, a better product, or pay 500 CZK, you have 10cm bigger TV or the same/similar product at a lower price, and so on.

Why implement it? Current issue:
The customer chooses the product and does not know that he can have a better product or product, including accessories, for similar money.

Benefit for our customer:
The customer gets a better product than the one he has chosen, and easier product comparison and selection.

Benefits for e-shop:
Increase revenue and order value.

Data and inputs:
Product Database at Mall.cz (XML Product Feed).

Supervisor:
Petr Němeček

User Profilling

What’s going on:
When sorting products in individual categories on our e-shops, no user preferences are currently taken into account. All customers always see the same results no matter what brands and price levels they favour or what they have previously made. The goal is to integrate product preferences into product sorting and thereby deliver to the user the products they expect and is willing to buy.
For example, if a customer searches for “laptops” at Mall.cz and we know he prefers laptops from Lenovo, and we know that the electronics “is not much like”, we prefer notebooks from Lenovo at a price of 15 to 20,000 CZK.
If a customer selects clothing at Mall.cz, we may prefer, for example, preferred sizes, style or colours.
The goal of this task is to build a user profile that will then be used in the search. Anonymized user data is used to create a profile.

Why implement it? Current issue:
The goal is that our customer will always see the items that are relevant to him in the first positions. More and more customers are using mobile phones for their purchases, where crawling less relevant products is annoying and filtering more demanding. Additionally, if a user buys regularly in a given category, he does not want to re-enter the same values.

Benefit for our customer:
In the first positions of the results, he will see his preferred products. He will not have to deal with a number of less interesting products. He will not be forced to use filters to further limit the number of products.

Benefits for e-shop:
Increase the number of orders.

Data and inputs:
Data capturing anonymized user behavior on the web (queries, purchases, …)

Supervisor:
Roman Dušek

Mentors

Extremely experienced mentors will be available 24 hours a day.

Radko Sekerka

Radko Sekerka

Head of Product @ Mall Group

He came from AVG where he managed online sales and monetization. With the right strategy, he has been able to massively drive sales across markets, especially in the US. In Mall Group, his mission is to improve the product to work and grow.

Petr Němeček

Petr Němeček

Head of Product Management @ Mall Group

Petr graduated from FIT BUT in Brno. He worked in AVG in various positions from the tester, through analytics to the Product Manager of AVG Antivirus. He went to Mall Group, where he held the role of Head of Product Management. It is part of a central product team that defines the development of the group’s e-shops. Petr is in charge of prioritizing and cross-referencing the product roadmap Eshop Mall for all countries where the Mall sells.

Roman Dušek

Roman Dušek

Head of Search @ Mall Group

Previously he worked as a database specialist in some startups. Later, he developed the Internet search in List.cz as a product manager. He focused primarily on automatic user satisfaction, analysis of the information needs and vertical search in structured and semi-structured data. For several years, he has been running a search team at Seznam.cz. He is currently working in the Mall Group, where he is in charge of a product search engine.

Michal Moravec

Michal Moravec

Product Manager @ Mall Group

Michal has been working on online projects for a long time. Previously, he was primarily concerned with online projects in the banking sector. Now in Mall Group, he manages projects in cooperation with external contractors, deals with areas of innovation, chatbot, mobile apps and more.

Ilja Volf

Ilja Volf

Head of Business Intelligence @ Mall Group

In the past, Ilja worked for a global corporation, a medium-sized regional firm and a startup. He has managed data projects and projects that have changed operations across logistics, manufacturing, IT and e-commerce. He built teams in Demand Generation, Research, BI and Analytics, Backoffice, and in technological development. And now, he builds a top class data team for MallGroup;)

Viktor Šohájek

Viktor Šohájek

Data Scientist @ Mall Group

Viktor studied Applied Mathematics at CTU, then worked as a BI developer for the E-commerce Holding + some small clients, from where he moved to Data Scientist at Mall Group. Now he is a part of a small team that mainly cares about predictive modelling – a lot of time is devoted to the dynamic pricing of all products, the forecasting of marketing KPIs, or the evaluation of marketing campaigns by modelling time series.

Jiří Riedl

Jiří Riedl

Head of Development @ Finnology

He started as a programmer and later as an application architect at systems for the financial market. Later he turned to strategic management. When he returned to lead developers, he tried to work in Telco and E-commerce, but eventually pulled back the financial market. Now he is working in Finnology. As an application architect and development manager is responsible for the solution architecture, which has ambitions to get into the whole world.

Petr Stýblo

Petr Stýblo

IBM Cloud

Petr designs and sells cloud software solutions. He devotes himself mostly to cloud-native applications – doing microservices, networking, DevOps, and often testing IoT or web technologies. He does not deal professionally with software development, but he is a Linux geek and, in his spare time, he tracks and supports a number of open-source projects and developer communities. He has also recently dealt with the theme of artificial intelligence, and he believes he can complicate the future dominance of machines.

Jiří Pětník

Jiří Pětník

IBM GTS IT Specialist

Eight years ago, he began to work as a trainee at IBM. He focused on managing end devices and cloud solutions. Currently, he works in the service department and is in charge of the development. He participates in projects for state administration, banks and insurance companies. As a member of the local IoT / AI team, he has long been focusing on designing solutions on the Internet for things not only using the IBM Bluemix cloud platform and IBM Watson cognitive services.

Tomáš Kadlec

Tomáš Kadlec

IBM Certified IT Architect

More than 20 years of experience in complex IT solutions. Practical experience with complex information systems in banking, insurance, healthcare and the public sector. He primarily performs the role of expert consultation in the pre-sales phase of the project, focusing on solutions utilizing IBM’s innovative products in the area of cognitive, cloud and mobile technologies. He has practical experience with developing mobile and web applications, iOS / Swift, Bluemix / Node.js and WebSphere / Java.

Registration

In addition to the personal participation of at least one team member at the Kick-off Meeting, a symbolic registration fee is required to participate in the hackathon. We take it as a signal that you really take part in the event and we can provide a place for you and enough refreshments. Its amount is only CZK 200 per person. By registering, you agree to the terms of the competition. You can find them below in the FAQ.

FAQ

What are the conditions of the competition?

We tried to make it as simple as possible.
By registering to the hackathon, you automatically agree with them.
You can find their full text here: http://bit.ly/RockawayHackathon-2017-HerniRad

Why we do that?

We enjoy problems that are difficult at first glance, but the technology can easily cope with them. The purpose of Rockaway Hackathon 2017 is to design an efficient and elegant solution using modern approaches.

Therefore, we are preparing the assignment to show the real and existing situations that our customers are addressing. We will look at real data that is not commonly available, and we expect a functional solution at the end – even so functional that we can integrate it into the Mall Group infrastructure.

When we make a long-term collaboration with the winning team, it will be a nice bonus for us.

BTW, the goal of the hackathon is to show the great potential of the IBM Bluemix platform.

Will there be a Kick-off Meeting?

Each registered participant will be invited to Slack Workspace in advance, where he will find mentors, sample data, and debates on topics.

Which technologies will be worked with?

Code in a language of your choice, there is no limit here. Because we love the latest technology, we will work with the IBM cloud service. You’ll find details on specific topics – but remember that you must be able to join the Mall Group infrastructure. We do not want to see competitive cloud services on the hackathon. 🙂

What will the winning teams get?

The winner takes it all.
Specifically, CZK 100,000.

Is it a problem if I don’t have my own team?

It is not! At the beginning, there will be an introductory presentation of the participants and their ideas. As an individual, you can join any existing team or create a part with others. The maximum number is 4 brains per group.

Why do I have to pay a registration fee?

Registration fee tells us that you are really serious about participating and we can reserve a chair for you. It also covers part of our costs for food and drinks.

How about sleeping?

There is a lot of space in the Hub – if you plan to sleep, we recommend you bring a blanket or sleeping bag.

Have we forgotten something?

Ask for anything

2015 & 2016

We have experience with hackathons - we have held them in Prague, Brno and Ostrava for the last 3 years, using AWS or Azure. We have always been astonished at how the teams have made a great atmosphere. And even better results.

Partners

These partners provide us with superior support.
Thank you!

Organizer