<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Vanessa Deborah Viellieber on Medium]]></title>
        <description><![CDATA[Stories by Vanessa Deborah Viellieber on Medium]]></description>
        <link>https://medium.com/@vviellieber?source=rss-21fc08dbf1a1------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*pmr3BNoCsC4OdWfrqAzPrA.jpeg</url>
            <title>Stories by Vanessa Deborah Viellieber on Medium</title>
            <link>https://medium.com/@vviellieber?source=rss-21fc08dbf1a1------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 21:09:13 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@vviellieber/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Best of Both Worlds: Data Science as a Supplement to Classical Research Methods]]></title>
            <link>https://medium.com/@vviellieber/the-best-of-both-worlds-data-science-as-a-supplement-to-classical-research-methods-c274ac6732ac?source=rss-21fc08dbf1a1------2</link>
            <guid isPermaLink="false">https://medium.com/p/c274ac6732ac</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[market-research]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[social-media]]></category>
            <category><![CDATA[nlp]]></category>
            <dc:creator><![CDATA[Vanessa Deborah Viellieber]]></dc:creator>
            <pubDate>Sun, 02 Jun 2019 09:56:30 GMT</pubDate>
            <atom:updated>2019-06-02T09:56:30.612Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A controversial discussion, illustrated by the mobility industry: Why the use of social media and other freely available data is necessary to ask the right questions in classical market research.</em></p><p>And another study on the subject of electromobility: meaningful composition of the sample; qualitative approach to the topic through a group discussion, followed by quantitative underpinning; numerous open and closed questions in online surveys with various choices, which means of transport the participants prefer and under which circumstances they could imagine buying an electric vehicle. The results? Not surprising for someone who is close to realising what is on everyone’s lips about electric mobility. Infrastructure, costs and charging time are the biggest obstacles for consumers to invest in an electric vehicle.</p><p>On Twitter, Facebook, Youtube, various blogs and specialist forums, mobile people share their mobility experiences and discuss the advantages and disadvantages of various products and services. By using their smartphones, e.g. to navigate or book train tickets, they leave a digital footprint or even a whole lane almost everywhere. By 2021, around 28 billion devices will be connected to the Internet. This opens up the potential to have a treasure trove of direct and personal information about customers. <br>If you take a look at the data extracted from the world wide web in terms of mobility behaviour, you come across one interesting aspect after another: no car sharing vehicle available within the next hundred meters, then an Uber; by bicycle to university, but only if it’s over six degrees and it’s not raining; on Saturdays by car instead of by train to yoga, because then a football match takes place again and the S-Bahn is unpleasantly crowded; with the car sharing vehicle, despite own vehicle in the garage, to the city, to avoid the annoying search for a parking space.</p><p>These kinds of interesting interrelationships are rarely encountered in relevant studies. They relate to the behaviour of people in special situations and can often only be identified with great effort using classical methods. It could be decisive for competition to know, <br>… that with a 20 percent higher area coverage the car sharing offer would be used 60 percent more; <br>… that customer satisfaction and loyalty will increase if the availability of public transport is increased in the event of bad weather forecasts or upcoming events with large crowds; <br>… that the abolition of public parking spaces will not reduce the volume of cars in the city centre.</p><p>In order to be competitive, it is more important than ever to develop business models whose products and/or services satisfy customer needs better than those of the competition. Above all, traditional companies are under great pressure to ensure their continued existence within industries that are undergoing major upheaval as a result of digitisation and in which new players are questioning market conditions. The automotive industry, for example, in which new value chains are emerging, old ones are breaking up and are increasingly developing into a diverse mobility industry.</p><h3>Interrelationships can be key competitive factors</h3><p>As Richard Branson, a successful investor and founder of the Virgin Group, which controls more than 400 companies, said:</p><blockquote>“Listen more than you talk!”.</blockquote><p>The development of social networks and digital platforms as well as the availability of other data, e.g. in open communities, increases the number of usable data and thus the potential of the associated opportunities to gain new insights. Customers, potential customers and anti-customers use various platforms on the Internet to exchange information about your product, your competitors’ products or their needs. Relevant information such as patterns, correlations, rules or relationships can be identified using data science analysis methods. Thus, the potential to learn business-relevant knowledge about your customers and competitors increases with the amount of data it is worth listening to.</p><p><strong>The challenge of data science analysis in terms of freely available data lies in the following areas </strong><br>(1) in the selection of the right data/ sources,<br>(2) in making the data available,<br>(3) in handling with the high number of data and<br>(4) the use of the correct methods of analysis in relation to the type of data.<br>Throughout the entire analysis process, it is essential to always have the actual research or analysis question as well as the analysis goal in mind. In an iterative procedure, potentially relevant data sources can be identified. The right search parameters make it possible to separate relevant data from irrelevant data. This corresponds to the process of information retrieval. Web scraping methods are used to extract the desired data from the network via appropriate interfaces.</p><p>In the following an overview of data, which can be opened up by appropriate analyses approximately:</p><p><strong>Demographic and psychographic data:</strong> Age, gender, income, marital status, employment status, education, attitudes, lifestyle, preferences, inclinations.<br><strong>Internet and mobile usage data: </strong>sessions, pages, referring sites, content displayed, navigation, transactions performed, app downloads, frequency of visits, length of stay, abandonment, feedback.<br><strong>Social media data: </strong>data on networks and influences, profiles, career development, group membership, product and company references (who likes whom is followed), online comments and customer reviews, customer voice.<br><strong>Signaler and sensor data:</strong> Varies by industry and application, typically including location, GPS, proximity and biometric data, RFID data and product sensor messages. In the mobility industry, for example, electric charging post status, particulate matter sensor data and traffic flow data.</p><h3>Processing of textual data</h3><p>80 percent of these data are available in unstructured form, and the trend is rising. Among the most important unstructured data available are texts, images, videos and voice recordings — practically all data that people generate in natural communication. They can be captured on a large scale, but special methods from data analytics are required to analyze them. A prototypical process of processing textual data is outlined in Figure 1.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qB2QvT1WOQRdDfMq1X9wxA.png" /><figcaption>Figure 1: From unstructured text data to structured form as input for further analysis</figcaption></figure><p>The goal of this process is to transform the data from an unstructured form into a structured form in order to identify patterns, relations and information contained in it. For this purpose, algorithms from the field of Natural Language Processing (NLP), i.e. the computer-based processing of natural language, are used. <br>Preprocessing is initially about data aggregation and cleansing. For example, texts from different users are merged and cleansed of unnecessary data variables. After the texts have been broken down into individual tokens (= words), the word type and negations within a text have been identified, interesting entities from the texts are identified in the fourth step, the Named Entity Recognition (NER). In abstract terms, this takes place in a two-step process. First, interesting terms are recognized as such. Then it is necessary to divide them into known classes of entities, e.g. products and functions of a product. With regard to ambiguities and the choice of the right meaning, it is necessary to be able to integrate domain knowledge into existing logic components. An example of an ambiguity of words is “Cayenne”: it could be the Cayenne pepper, the city on the French Polynesia or the Porsche vehicle model. The following steps are particularly relevant in the analysis of social media data in order, for example, to be able to infer from the context which products/ services the customers have evaluated positively or negatively.</p><p>Now that sufficiently structured information has been generated from text data, appropriate models can be developed. Sentiment analysis is a common method, especially in the social media analytics environment. This captures the mood within texts. If this is combined, for example, with a Topic Modeling, i.e. the determination of topics which are discussed within the texts, one is able to recognize which topics are written about which with which mood. <br>The increasing amount of available data in combination with the decreasing costs of computing power enable fast progress for deep learning procedures in the field of artificial intelligence (AI). This means that especially complex data processing methods, such as those for natural language processing, benefit from ever better methods and algorithms, so that high-quality results can increasingly be achieved more easily.</p><h3>Linkage with classical methods of market research</h3><p>One of the limitations of free data analysis is that it is hardly possible, if at all, to validate ground truth in online data, as there is no clear way to determine the accuracy of a person’s demographic information and the validity of contributions. This, in turn, involves a link with classical market research methods, which can be used to validate results and place them on a representative basis. In addition, it makes sense to examine the findings obtained with data science methods with the help of classical market research methods, since often only the actual behaviour is analysed and, as a result, the insights into the consumer’s world of thought may be neglected. In addition, depending on the data source, only digitally-affine and thus younger target groups tend to be the object of analysis.</p><p>Figure 2 illustrates the extent to which data science analyses and classical market research complement each other and thus combine new, statistically relevant perspectives that enable a holistic view of the topic. However, it cannot be ruled out that data science analyses may not be well-founded or that classical market research may not be explorative.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TTgw7hngvh4q25toxV_lJw.png" /><figcaption>Figure 2: Data science analyses and market Research: a complementary view</figcaption></figure><p>The result is a new model of interdisciplinarity. One possibility of a collaboration model is shown in figure 3. Once the analysis goal has been defined together with the experts, it is transferred to explorative data science analysis. Initial findings are used to select additional interesting data sources and make them accessible for the interpretation of the analysis results.</p><p>If interesting correlations, patterns or relationships can be found on the basis of which changes in strategy, business model, product features or marketing measures can be justified, or if one has simply reached the limits of the possibilities of data science analyses, the results must be examined more closely within the framework of classical market research. Findings from market research can justify new data science analyses. If the results serve the object of investigation, they have to be converted into corresponding measures downstream.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-ieWWjMqkcmo2F8r1f15zw.png" /><figcaption>Figure 3: Process mapping of the method combination of data science and classical market research</figcaption></figure><p>Thus, the findings from the open source data analyses broaden the perspective and enable on the one hand the definition of valid questions and on the other hand, due to a sharpened view or an extended perspective of the object of observation, a better interpretability of the results. This extended perspective and the accompanying ability to interpret can be further extended by linking the open source data with the exclusive company-internal data and thus uncovering latently existing patterns and interactions.</p><p>Have you already gained experience with the combination of data science and classical market research methods? Do you see it differently? I am interested in your opinion and look forward to a stimulating exchange.</p><p><em>First publication: 29.05.2019 on </em><a href="https://www.marktforschung.de/dossiers/themendossiers/text-und-voice/dossier/das-beste-aus-beiden-welten-data-science-als-ergaenzung-zu-klassischen-forschungsmethoden/"><em>marktforschung.de</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c274ac6732ac" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>