The rapid dissemination of the Internet and related services, especially social networks and user-generated content, led to incredible amounts of digital information, better known as big data. This, at the same time, resulted in both, a radical change and new opportunities in marketing.

Analysis of huge amounts of data requires special tools and methods. Industry managers and researchers increasingly embrace technological innovations that make decision-making effective in the context of big data.

Each of the three topics is contained in a separate block, which starts with a lecture covering the theoretical foundations, followed by an applied part. Your learning progress is evaluated by means of assignments, presentations, and a small multiple-choice test at the beginning of each unit covering the material from previous the session.

Although all slides, teaching materials and syllabus are offered in English, the lectures themselves are held exclusively in German. Therefore, basic knowledge of German is required.

The aim of this course is to provide students with an understanding of how big data can be used for customer analytics (customer demographics and behavioral analysis), contextual marketing (creating personalized recommendations), and online communication management (e.g., reputation management by using “shitstorm”-detection).

Students will learn how to use modern technologies for purposes like sentiment analysis, personalization, and data exploration to support managerial decision making and optimize customer satisfaction in a data-driven world.

After the course, you will be able to:

• Use common social media application interfaces for scraping purposes.
• Describe the benefits and limitations of sentiment analysis and recommendation engines.
• Use R for basic data exploration and visualization (dplyr & ggplot2).
• Build a basic social media crisis detection system based on state-of-the-art text mining technologies.
• Gain a better understanding of your customers by using text mining tools (e.g.,analyze unstructured feedback, complains or wishes). 

Class attendance is compulsory. Attendance of <80% of all sessions will lead to not passing the course.

The assignments may include the following tasks:

Programming tasks: Students have to adapt a given source-code template. Example: You get a template for scraping data from the twitter API and you have to adapt it to load only data from a specific user or hashtag.

Recherche tasks: The course covers only the basics of the introduced topics. So some tasks allow you to dig deeper into the topics. Example: You have to find out which types of recommendation systems are used by Netflix and Amazon.

Error detection tasks: You get a written text about one of the introduced topics or the implementation of an algorithm, which contains some errors. You will have to find and fix them. Example: You get an incorrect implementation of the Pearson correlation and you have to fix it!

Practical usage tasks: 90% of big data is unstructured (e.g., tweets, Facebook posts, images or videos). So statistical methods are used to analyze them automatically, but these approaches have limitations. For example they are not able to recognize sarcasm. Example: One task could be to write a sarcastic tweet and “trick” a program to assign the wrong sentiment to it.

Answering questions: You have to provide the right answer given a question. In some cases a few options are already given (multiple choice).

• You can obtain max. 10 points for each assignment (30 points in total).
• Students have one week to complete the tasks in small groups (2-3 people).
• Please note that submissions after the deadline will not be graded.

Grading is based on the following components:

These grading components will be added to calculate the final grade, which is based on the following grading scheme:

To successfully pass this course, your cumulated points need to exceed 60 points.

Please note that to ensure an equal contribution of group members for all group assignments, a peer assessment will be conducted among group members, which enters into the computation of the individual grades for your group work. This means that the members of a group are required to assess other students regarding their relative contribution.

Assignments

• There is one applied assignment foreach of the three blocks.
• Students require a basic understanding of the blocks’content to solve the respective assignments.

I am happy to answer your questions. So feel free tosend me a short email or drop by my office after making an appointment if you would like to talk to me inperson. I will also try to be available in the classroom after each class orduring the breaks of each class.

Recommended readingfor the course:

Data Exploration

• http://r4ds.had.co.nz/

Recommendation Systems

• http://guidetodatamining.com/
• Bell, Robert M., andYehuda Koren. "Lessons from the Netflix prize challenge." ACM SIGKDD Explorations Newsletter 9.2 (2007): 75-79.
• Koenigstein, Noam, Gideon Dror, and Yehuda Koren. "Yahoo! musicrecommendations: modeling music ratings with temporal dynamics and itemtaxonomy." Proceedings ofthe fifth ACM conference on Recommender systems. ACM, 2011.

Text Mining

• https://www.r-bloggers.com/sentiment-analysis-with-machine-learning-in-r/
• Godbole, Namrata, ManjaSrinivasaiah, and Steven Skiena. "Large-Scale Sentiment Analysis for Newsand Blogs." ICWSM 7.21 (2007):219-222.
• Pak, Alexander, andPatrick Paroubek. "Twitter as a Corpus for Sentiment Analysis and OpinionMining." LREc. Vol. 10.2010.
• Kouloumpis, Efthymios,Theresa Wilson, and Johanna D. Moore. "Twitter sentiment analysis: Thegood the bad and the omg!." Icwsm 11 (2011):538-541.

The courseslides will be made available prior to class via Learn@WU. Please make surethat you have access to all course materials! If not, please contact me!

This course provides a deeper insight into the following three data-driven contributions to marketing and sales, enabled by social media and modern big data technologies:

Customer Analytics: As a recent study revealed that customer analytics are dominating big data use in sales and marketing (Datameer 2015). Among other things, it aims to improve conversion rates and lower customer acquisition costs. In this course, R (a well-known and free statistical programming language) will be used to analyze customer demographics and behavior. To provide an adequate setting, a real-world data set will be investigated in detail.

Contextual Marketing: Social media platforms know a lot about their customers by gathering huge amounts of data. For example, the video streaming service Netflix uses this information in combination with a personalized recommendation engine to reach high customer satisfaction. In this course, a consumer database will be used to build such a recommendation system. The engine will be based on collaborative filtering, a method of making automatic predictions of the interests of users based on users’ preferences.

Online Communication Management: Social networks have revolutionized communication, providing a public channel where customers can contact businesses, organizations or people directly. On the one hand, these channels allow companies to gain a better understanding of the general market and of their own as well as their competitors' customers. On the other hand, they also pose a threat to any company’s reputation, because they may be used for customer attacks. Nowadays bigger companies are prepared for social media crises (also known as “shitstorms”) by using modern detection technologies. In this part of the course, students will have to create a small scraper, which loads data from a so-called social media application programming interface (API). In a second stage, state-of-the-art text mining techniques will be used to determine the sentiment of the collected data.