<?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 Khanimazimli on Medium]]></title>
        <description><![CDATA[Stories by Khanimazimli on Medium]]></description>
        <link>https://medium.com/@khanimazimli0?source=rss-e105e927c97------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*TIHHDeBoEQekKVKOiGzasA.jpeg</url>
            <title>Stories by Khanimazimli on Medium</title>
            <link>https://medium.com/@khanimazimli0?source=rss-e105e927c97------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 03:27:43 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@khanimazimli0/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[Multiple choice questions are often treated as the simplest part of a survey.]]></title>
            <link>https://medium.com/@khanimazimli0/from-checkbox-answers-to-customer-insight-from-percentages-to-behavioral-models-with-python-21599842c658?source=rss-e105e927c97------2</link>
            <guid isPermaLink="false">https://medium.com/p/21599842c658</guid>
            <dc:creator><![CDATA[Khanimazimli]]></dc:creator>
            <pubDate>Mon, 06 Apr 2026 08:08:12 GMT</pubDate>
            <atom:updated>2026-04-07T06:08:42.537Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><strong><em>Beyond Percentages: Turning Multiple Choice Answers into Behavioral Models with Python</em></strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ikrv25SpmBSMRdWk" /></figure><p>Multiple choice questions are often treated as the simplest part of a survey.</p><p>They are easy to ask, easy to answer, and easy to visualize.</p><p>And yet -in most CX and product teams -they are one of the most <strong>misinterpreted</strong> data types.</p><p>Because we reduce them to percentages.</p><blockquote>45% selected Payments<br>32% selected Transfers</blockquote><p>And we stop there.</p><p>But multiple choice questions are not just distributions. They are <strong>multi-dimensional behavioral signals</strong>.</p><p><strong><em>The main problem: Flattening Behavioral Data</em></strong></p><p>A multiple choice question is not a single variable.</p><p>Each respondent generates:</p><ul><li>multiple binary signals</li><li>overlapping preferences</li><li>implicit behavioral patterns</li></ul><p>When we say:</p><p><em>“45% selected Payments”</em></p><p>we <strong><em>ignore</em></strong>:</p><ul><li>what else those users selected</li><li>whether they behave similarly</li><li>whether that 45% represents one segment or multiple distinct ones</li></ul><p><strong>We flatten multi-dimensional behavior into single-dimensional metrics.</strong></p><p><strong>Step 1: Structuring the Data</strong></p><p>Most datasets arrive as raw strings.</p><p><em>df[“usage”].str.get_dummies(sep=”, “)</em></p><p>This transforms responses into a <strong>binary feature matrix</strong>.</p><p>At this point, we are no longer working with survey answers.<br>We are working with a <strong>behavioral dataset</strong>.</p><p><strong>Step 2: Frequency ≠ Insight</strong></p><p><em>dummies.sum()</em></p><p>This answers:</p><p>What is popular?</p><p>But it does NOT answer:</p><ul><li>what behaviors co-exist</li><li>what differentiates users</li><li>what defines segments</li></ul><p><strong>Frequency is descriptive -not explanatory.</strong></p><p><strong>Step 3: Co-Occurrence- With a Caveat</strong></p><p><em>co_occurrence = dummies.T.dot(dummies)</em></p><p>This shows which features appear together.</p><p>But raw co-occurrence has a limitation:</p><p><em>Popular features will appear correlated with everything.</em></p><p>To understand whether relationships are meaningful, we need normalization:</p><ul><li>Jaccard similarity</li><li>cosine similarity</li><li>lift (association rules)</li></ul><p>These help answer:</p><p>“Is this relationship structurally significant, or just frequent?”</p><p><strong>Step 4: From Co-Occurrence to Structure</strong></p><p>At this point, many analyses jump to intuitive clustering.</p><p>For example:</p><blockquote>Payments + Transfers → “transactional users”</blockquote><blockquote>Savings + Investments → “planning users”</blockquote><p>This is useful, but only as a <strong>hypothesis, not a model</strong>.</p><p><em>To formalize this, we need:</em></p><blockquote>clustering (k-means, hierarchical)</blockquote><blockquote>association rules (Apriori, FP-Growth)</blockquote><p>Without this step, insights remain interpretative, not validated.</p><p><strong>Step 5: The Combinatorial Reality</strong></p><p>In small datasets, combinations are manageable.</p><p>But in real-world surveys:</p><blockquote>20+ options</blockquote><blockquote>thousands of users</blockquote><p><strong>combinations grow exponentially</strong></p><p>This creates:</p><ul><li>noise</li><li>sparsity</li><li>interpretability challenges</li></ul><p>Which means:<br><em>we must prioritize patterns, not enumerate everything</em>.</p><p><strong>Step 6: Adding Context: The Missing Layer</strong></p><p>Patterns alone are not insight.</p><p>Example:</p><blockquote>Payments + Transfers</blockquote><p>This pattern means very little by itself.</p><p>Insight emerges only when combined with:</p><ul><li>income</li><li>activity level</li><li>transaction volume</li><li>lifecycle stage</li></ul><p>Without this, interpretations remain speculative.</p><p><strong>Mini Case: From Survey Answers to Actionable Segments</strong></p><p>Let’s move from theory to a simplified example.</p><p>Assume we asked:</p><p><strong>“What do you primarily use your banking app for?”</strong></p><p>We transform responses into a binary matrix (dummies) and apply association rule mining.</p><p><strong>Step 1: Finding Frequent Patterns</strong></p><blockquote>from mlxtend.frequent_patterns import apriori, association_rules</blockquote><blockquote>frequent_itemsets = apriori(dummies, min_support=0.2, use_colnames=True)<br>rules = association_rules(frequent_itemsets, metric=”lift”, min_threshold=1.2)</blockquote><blockquote>rules[[“antecedents”, “consequents”, “support”, “confidence”, “lift”]]</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AvqRqXlkdfT4QcxqExSsTQ.png" /></figure><p><strong>What This Means</strong></p><p>Now we can say:</p><ul><li>Users selecting <strong>Savings</strong> are <strong>2.1x more likely</strong> to also select Investments</li><li>Payments → Transfers relationship is structurally strong, not just frequent</li></ul><p>Importantly:</p><p><strong>lift &gt; 1 indicates a positive association</strong><br>lift ≈ 1 → no meaningful relationship<br>lift &lt; 1 → negative association</p><p>This is the difference between:</p><ul><li>raw co-occurrence</li><li>and <strong>validated behavioral association</strong></li></ul><p><strong>Step 2: Translating into CX Segments</strong></p><p>Instead of overclaiming:</p><p>❌ “These are investment users”</p><p>We say:</p><p><em>“Users exhibiting a Savings + Investments pattern show a strong behavioral association, indicating potential long-term financial intent”</em></p><p>This shift increases analytical credibility.</p><p><strong>Step 3: Business Impact</strong></p><p>This analysis directly informs decisions:</p><p><strong>Personalization</strong></p><p>Payments + Transfers<br>→ speed, shortcuts, friction reduction</p><p>Savings → Investments<br>→ education, long-term value messaging</p><p><strong>Campaign Targeting</strong></p><p>Instead of:</p><blockquote><em>“Target users interested in savings”</em></blockquote><p>You define:</p><p>Users with:</p><ul><li>Savings behavior</li><li>high association to Investments</li><li>but no actual investment usage</li></ul><p>This becomes a <strong>conversion opportunity segment</strong>.</p><p><strong>Product &amp; Revenue Impact</strong></p><ul><li>Better targeting → higher conversion</li><li>Better segmentation → reduced irrelevant communication</li><li>Better insight → smarter feature prioritization</li></ul><p>Example:</p><p>Strong Savings → Investments relationship<br>→ investment onboarding friction becomes a high-impact product problem</p><p><strong>Step 4: Making It Visible</strong></p><blockquote>import seaborn as sns<br>import matplotlib.pyplot as plt</blockquote><blockquote>sns.heatmap(dummies.T.dot(dummies), annot=True, fmt=&quot;d&quot;)<br>plt.title(&quot;Feature Co-Occurrence Matrix&quot;)<br>plt.show()</blockquote><p>Even simple visuals help reveal:</p><ul><li>behavioral clusters</li><li>structural relationships</li></ul><p>⚠️ Real-World Caveat</p><p>Multiple choice data has limitations.</p><ul><li>It is <strong>self-reported</strong>, not observed behavior</li><li>Users may select aspirational answers</li><li>Survey context can influence responses</li></ul><p>This means:</p><p><strong>survey data should ideally be validated with behavioral data (transactions, usage logs)</strong></p><p>Final</p><p>Multiple choice questions are not simple.</p><p>They are:<br><em>compressed behavioral datasets</em></p><p>Python helps structure and explore them.</p><p>But insight comes from:</p><p><strong><em>combining methods, context, and interpretation</em></strong></p><p>***</p><blockquote><em>The real value of multiple choice questions is not in what users select-<br>but in what their selections reveal when modeled, validated, and connected to behavior.</em></blockquote><blockquote><em>And that is where survey analysis ends, and customer understanding begins.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=21599842c658" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Can Research Enter Your Life While You’re Still a Child? Yes. And It Can Prepare You for Life Too.]]></title>
            <link>https://medium.com/@khanimazimli0/can-research-enter-your-life-while-youre-still-a-child-yes-and-it-can-prepare-you-for-life-too-f3c48f821c02?source=rss-e105e927c97------2</link>
            <guid isPermaLink="false">https://medium.com/p/f3c48f821c02</guid>
            <category><![CDATA[human-centered-design]]></category>
            <category><![CDATA[ux-research]]></category>
            <category><![CDATA[inquiry-based-learning]]></category>
            <category><![CDATA[ib-education]]></category>
            <dc:creator><![CDATA[Khanimazimli]]></dc:creator>
            <pubDate>Sun, 23 Mar 2025 16:13:37 GMT</pubDate>
            <atom:updated>2025-03-23T16:13:37.076Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*c7AV_RBJm_fJTC2y7m2s1Q.jpeg" /></figure><p>I worked for nearly five years at TISA, a school that offers the International Baccalaureate (IB) programme. During that time, I witnessed firsthand how powerful it is when education and research come together. Even at the primary level, IB students are introduced to the simplest — but most meaningful — forms of inquiry.</p><p>For example, at the start of the school year, 5th graders are asked to choose a topic. Throughout the term, they research this topic using different methods — reading, interviews, experiments — and finally, present it at an exhibition open to parents, teachers, and peers. From early on, these children are not just memorizing facts. They are learning how to <em>learn</em> through research. They ask questions, seek answers, and develop critical thinking.</p><h3>What Does Research Say About IB Students?</h3><p>IB students are known for their ability to think critically, seek out primary sources, and constantly question and challenge information. They develop an interdisciplinary mindset that helps them understand and contribute to solving global issues. IB learners show leadership, empathy, and personal growth through community engagement. They grasp the interconnectedness of the world and the importance of international cooperation (Paris, Paul. (2003))</p><p>They also appreciate diverse perspectives — something they experience not only at home or while traveling, but through multilingual learning and digital interactions. These students speak multiple languages, which broadens their worldview and enables them to understand different thought processes. According to the IB Organization, students who complete the IB Diploma Programme are mature, self-aware, and ready to succeed in higher education and beyond, armed with strong values, cultural sensitivity, and confidence.</p><h3>So How Does This Relate to UX Research?</h3><p>Today, I work as a UX researcher. And surprisingly — or maybe not so surprisingly — I find strong parallels between the IB approach and my current role. In UX, we aim to understand real people, their behaviors, needs, and emotions. We ask questions. We observe. We challenge assumptions. We make sense of data. And most importantly, we reflect.</p><p>These are the exact same skills I saw being nurtured in IB classrooms. The process of inquiry, of being curious and evidence-based, is not something that starts in adulthood. It can begin as early as elementary school.</p><p>As UX researchers, our goal is to design better experiences by listening deeply and thinking critically. But imagine a generation that grows up already trained to do just that. That’s what the IB programme offers — an education rooted in research, reflection, and relevance.</p><h3>Final Thoughts</h3><p>Research isn’t just a skill — it’s a mindset. And that mindset can be developed early in life. The IB programme fosters that beautifully. I’ve witnessed children becoming active inquirers, and now I apply the same mindset in my job every day. The connection between the two is clear: whether in school or in product development, those who learn <em>how</em> to ask, explore, and reflect are the ones best prepared for the complexities of the real world.</p><p>References:<br>Paris, Paul. (2003). The International Baccalaureate: A Case Study on why Students Choose to do the IB. 4.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f3c48f821c02" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>