{"id":59390,"date":"2025-01-31T06:31:03","date_gmt":"2025-01-31T06:31:03","guid":{"rendered":"https:\/\/peraltafinancing.com\/analytics\/how-do-you-use-sequential-data-to-refine-navigational-analysis%e2%9d%93\/"},"modified":"2025-01-31T06:31:03","modified_gmt":"2025-01-31T06:31:03","slug":"how-do-you-use-sequential-data-to-refine-navigational-analysis%e2%9d%93","status":"publish","type":"post","link":"https:\/\/fivemor.com\/?p=59390","title":{"rendered":"How do you use sequential data to refine navigational analysis\u2753"},"content":{"rendered":"<p> <br \/>\n<\/p>\n<div id=\"wtr-content\" data-bg=\"#0cacf8\" data-fg=\"#0cacf8\" data-width=\"5\" data-mute=\"\" data-fgopacity=\"1.00\" data-mutedopacity=\"1.00\" data-placement=\"top\" data-placement-offset=\"48\" data-content-offset=\"0\" data-placement-touch=\"top\" data-placement-offset-touch=\"0\" data-transparent=\"1\" data-shadow=\"1\" data-touch=\"\" data-non-touch=\"1\" data-comments=\"0\" data-commentsbg=\"#0cacf8\" data-location=\"page\" data-mutedfg=\"#0cacf8\" data-endfg=\"#f44813\" data-rtl=\"\">\n<p>When it comes to analysing behavioural web analytics data, analysts often work with aggregates: visit duration, number of page views, bounce rate\u2026 These variables make it possible to describe each navigation sequence in a very structured way. They are easy to represent for analysis purposes, or to use in a machine learning environment for example.\u00a0<\/p>\n<p>In this article, we provide suggestions for processing navigation data at a somewhat more refined, less aggregated scale, and with new information. The aim is to exploit the sequences of events constituting the visits in a quantitative and fully automated way.\u00a0More specifically, we look at page sequences over the course of visits, bearing in mind that the approach is transposable to sequences of events, products, etc.\u00a0<\/p>\n<p>The first step is to define the objectives of sequence data processing and then to outline a procedure for the automatic extraction of sequences of interest.\u00a0<\/p>\n<h2><strong>Objectives: identify and exploit sub-sequences of interest<\/strong>\u00a0<\/h2>\n<p>As\u00a0<a href=\"https:\/\/blog.atinternet.com\/en\/data-science-is-descriptive-predictive-and-even-prescriptive-j-bureau-data-scientist\/\" target=\"_blank\" rel=\"noreferrer noopener\">Data Scientists at AT Internet<\/a>, our role is to design solutions that automate the extraction of information and assist our users in their analyses. Since our data model allows us to define objectives (financial\u00a0or not) on websites, we decided to use it as a variable of interest. In this context, we aim to:\u00a0<\/p>\n<ul>\n<li>Identify navigation sub-sequences in relation to the conversion rate and highlight these relationships.\u00a0\u00a0<\/li>\n<li>Annotate the presence of such sub-sequences in the customer journeys and use them as predictive variables in a machine learning model.\u00a0<\/li>\n<\/ul>\n<p>For example: a website records a conversion rate of 3%, but we find that visits to pages A, B and then C generate a conversion rate of 14%. The sequence A&gt;B&gt;C has an associated conversion rate higher than the average conversion rate. It can therefore be considered as a sub-sequence of interest.\u00a0<\/p>\n<h2><strong>Data sets and modelling<\/strong>\u00a0<\/h2>\n<h3><strong>Identifying data sets<\/strong>\u00a0<\/h3>\n<p>First of all, we need to give some context to our navigation sequence data:\u00a0<\/p>\n<ul>\n<li>Each\u00a0sequence\u00a0is annotated as successful or unsuccessful.\u00a0<\/li>\n<li>We work with sequences of pages visited, for\u00a0example :\u00a0[Home &gt; Product1 &gt; Conditions of sale]\u00a0<em>(the sequences are deliberately reduced when the conversion pages are reached in order not to introduce a bias).<\/em>\u00a0<\/li>\n<\/ul>\n<p>In the end, you get a dataset like this one:\u00a0<\/p>\n<figure class=\"wp-block-table\">\n<table>\n<tbody>\n<tr>\n<td>user\u00a0<\/td>\n<td>sequence\u00a0<\/td>\n<td>conversion\u00a0<\/td>\n<\/tr>\n<tr>\n<td>123\u00a0<\/td>\n<td>Homepage &gt; Product1 &gt; Privacy &gt; Contact\u00a0<\/td>\n<td>false\u00a0<\/td>\n<\/tr>\n<tr>\n<td>456\u00a0<\/td>\n<td>Product 1 &gt; Product 2 &gt; Product 1 &gt; Return Policy\u00a0<\/td>\n<td>true\u00a0<\/td>\n<\/tr>\n<tr>\n<td>789\u00a0<\/td>\n<td>Product 2\u00a0<\/td>\n<td>false\u00a0\u00a0<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n<p>Such data sets are often\u00a0<em>im<\/em><em>balanced<\/em>. The group that does not convert is often in the majority: they frequently represent more than 95% of visitors. This is a problem frequently encountered in the\u00a0machine\u00a0learning process (conversion, fraud detection, churn, \u2026) and which could be the subject of an entire article by itself!\u00a0<\/p>\n<h2><strong>Modelling and evaluation of sequences of interest<\/strong>\u00a0<\/h2>\n<h3><strong>What are the objectives?<\/strong>\u00a0<\/h3>\n<p>A sequence of interest is defined by two\u00a0parameters :\u00a0\u00a0\u00a0<\/p>\n<ul>\n<li>Its\u00a0presence must affect the conversion rate (positively or negatively).\u00a0<\/li>\n<li>It\u00a0must be sufficiently present for its impact to be significant across the dataset.\u00a0<\/li>\n<\/ul>\n<h3><strong>Quantitative evaluation<\/strong>\u00a0<\/h3>\n<p>In science, it is best practice to reason quantitatively, so we\u2019ll\u00a0start\u00a0by focusing on a metric that will be used to evaluate potential candidate navigation sequences. In this way, we will be able to classify them and retain only the most useful ones.\u00a0<\/p>\n<p>In information theory, there is a metric that quantifies uncertainty: entropy. If we consider Y\u2208{0,1} a random variable indicating whether there has been a conversion, the entropy H(Y) makes it possible to quantify how random this event is on a scale of 0 to 1. In the case of a binary variable, the entropy is simply defined as follows:\u00a0<\/p>\n<p class=\"has-text-align-center\">H(Y)=\u2212(P(Y=0)log(P(Y=0))+P(Y=1)log(P(Y=1)))\u00a0<\/p>\n<p>(1)\u00a0<\/p>\n<p>or\u00a0P(Y=1)\u00a0which means: \u201cthe probability that Y is equal to 1\u201d, i.e. the probability that a conversion will take place. We can estimate this probability by the observed conversion rate.\u00a0<\/p>\n<p>A simple graph helps\u00a0to interpret this:\u00a0<\/p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" loading=\"lazy\" width=\"562\" height=\"301\" src=\"https:\/\/blog.atinternet.com\/wp-content\/uploads\/2020\/09\/image.png\" alt=\"\" class=\"wp-image-18005\" srcset=\"https:\/\/blog.atinternet.com\/wp-content\/uploads\/2020\/09\/image.png 562w, https:\/\/blog.atinternet.com\/wp-content\/uploads\/2020\/09\/image-300x161.png 300w\" sizes=\"auto, (max-width: 562px) 100vw, 562px\"\/><\/figure>\n<p>The further the probability of the outcome deviates from equilibrium (0.5), the less uncertainty there is about the outcome (so far, logical!).\u00a0<\/p>\n<p>What interests us is the impact of the presence of any sequence of pages (e.g. A&gt;B&gt;C) on the realisation of Y. By noting\u00a0X\u00a0the presence of A&gt;B&gt;C in the sequence (1 if present, 0 if not), we can define the\u00a0<strong>conditional entropy<\/strong>:\u00a0<\/p>\n<p class=\"has-text-align-center\">H(Y|X)=P(X=1)\u2005H(Y|X=1)\u2005+\u2005P(X=0)\u2005H(Y|X=0)\u00a0<\/p>\n<p>(2)\u00a0<\/p>\n<p>or\u00a0H(Y|X=1)\u00a0which means:\u00a0\u201cthe entropy of Y, when X is 1\u201d, i.e. the uncertainty that reigns over the act of conversion, when we know that the sequence A&gt;B&gt;C is included in the sequence of pages visited.\u00a0<\/p>\n<p>Conditional entropy now allows us to quantify the uncertainty on the outcome of\u00a0Y when\u00a0the outcome of X is known.\u00a0<\/p>\n<p>We have now brought together the two ingredients needed to calculate\u00a0<strong>mutual\u00a0information<\/strong>:\u00a0<\/p>\n<p class=\"has-text-align-center\">I(X,Y)=H(Y)\u2212H(Y|X)\u00a0<\/p>\n<p>(3)\u00a0<\/p>\n<p>We therefore read that\u00a0I(X,Y)\u00a0is the difference between the uncertainty of the outcome of Y (will the user convert?) and the uncertainty of\u00a0Y\u00a0<em>when<\/em>\u00a0X\u00a0is\u00a0<em>known<\/em>\u00a0(will the user convert\u00a0<em>knowing that they have carried out the sequence A&gt;B&gt;C<\/em>?).\u00a0So,\u00a0this is precisely the\u00a0<em>information<\/em>\u00a0that\u00a0X<em>\u00a0<\/em>carries about\u00a0Y.\u00a0<\/p>\n<p>It is important to note that in addition to meeting our need for information extraction, mutual information also depends on the frequency of appearance of sequences (see\u00a02) and the conversion rate when they are present or not (see\u00a01 and 2), as desired in the previous section.\u00a0<\/p>\n<p><em>Information theory was founded by Claude Shannon in the late 1940s and was initially applied in the field of telecommunications. Today it is present in many algorithms used for artificial intelligence applications. Its foundations are presented here in a very simplified way, but a solid mathematical theory guarantees the concepts presented. This\u00a0<\/em><a href=\"https:\/\/www.youtube.com\/watch?v=Ebx8GKBBya4\" target=\"_blank\" rel=\"noreferrer noopener\"><em>video<\/em><\/a><em>\u00a0<\/em><em>(in French with subtitles)\u00a0<\/em><em>details the origins of this exciting field.<\/em>\u00a0<\/p>\n<h3><strong>Candidate sequences<\/strong>\u00a0<\/h3>\n<p>Now that we are fully equipped to classify sequences by utility, we need to define a strategy for generating candidate sequences. The search space is indeed rapidly becoming vast: if we take a website with 15 pages and consider only navigation sequences of 20 pages at most, we\u00a0have to\u00a0test more than\u00a015^20\u22483.10^23\u00a0different sequences, which is\u00a0about the number of stars in the observable universe \u2013 quite a few\u2026\u00a0<\/p>\n<p>The choice of our generation of candidate sequences is guided by two needs:\u00a0\u00a0\u00a0<\/p>\n<ul>\n<li>The\u00a0generation must be based on simple algorithms, leaving room for future iterations and being able to be presented to less technical\u00a0stakeholders.\u00a0<\/li>\n<li>It\u00a0must use algorithms available in\u00a0<em>big data<\/em>\u00a0frameworks, in order to guarantee its generalisation to all AT Internet clients, whose data volumes vary considerably.\u00a0<\/li>\n<\/ul>\n<p>For these two reasons we have chosen the\u00a0PrefixSpan\u00a0algorithm, which can be explained in a few lines with a schema and whose implementation is available in the\u00a0<a href=\"https:\/\/spark.apache.org\/mllib\/\" target=\"_blank\" rel=\"noreferrer noopener\">Apache Spark MLlib<\/a>.\u00a0<\/p>\n<p>PrefixSpan\u00a0is a sequential pattern mining algorithm: in our sequence dataset, it is about identifying the most frequent sub-sequences. The algorithm has two parameters:\u00a0<\/p>\n<ul>\n<li>The\u00a0<em>minimum support<\/em>: the minimum frequency of a sub-sequence through the dataset to be considered frequent.\u00a0<\/li>\n<li>The\u00a0maximum length of the sub-sequences searched for, in order to limit exploration.\u00a0<\/li>\n<\/ul>\n<p>PrefixSpan\u00a0guarantees the extraction of all frequent sub-sequences (in the sense of\u00a0<em>minimum support<\/em>) \u2013 for this purpose it explores the tree of possible\u00a0sub-sequences. The diagram below is an example with pages named a, b, c, \u2026, z. A path from the root of\u00a0the tree to a node therefore represents a sub-sequence of visited pages.\u00a0<\/p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" loading=\"lazy\" width=\"622\" height=\"445\" src=\"https:\/\/blog.atinternet.com\/wp-content\/uploads\/2020\/10\/prefixspan_graph.png\" alt=\"\" class=\"wp-image-18112\" srcset=\"https:\/\/blog.atinternet.com\/wp-content\/uploads\/2020\/10\/prefixspan_graph.png 622w, https:\/\/blog.atinternet.com\/wp-content\/uploads\/2020\/10\/prefixspan_graph-300x215.png 300w\" sizes=\"auto, (max-width: 622px) 100vw, 622px\"\/><\/figure>\n<p>As mentioned above, it is unthinkable to explore this tree entirely. This is where\u00a0PrefixSpan\u00a0comes into play \u2013 by reducing the cost of exploration using two strategies:\u00a0<\/p>\n<ul>\n<li><strong>Pruning\u00a0<\/strong><strong>of the tree by the\u00a0<\/strong><strong><em>anti-monotonicity<\/em><\/strong><strong>\u00a0rule<\/strong>: if the A&gt;B&gt;C sub-sequence does not reach the frequency threshold, any sequence of which it is the prefix will have a lower frequency. For\u00a0example\u00a0A&gt;B&gt;C&gt;D will necessarily be less frequent than A&gt;B&gt;C, since the\u00a0second\u00a0is included in the\u00a0first. There is therefore no need to explore the sequences for which A&gt;B&gt;C is the prefix, so we can prune all the branches that follow and reduce the calculation.\u00a0<\/li>\n<li><strong>The\u00a0<\/strong><strong>projection of candidate post-fixes<\/strong>: at each step, only the post-fixes (sequences following the prefix) still candidate for the calculation are kept. For example, if we have just processed the sequence A&gt;B&gt;C, we have eliminated at the first stage all the\u00a0sub-sequences not beginning with A, then all those not beginning with A&gt;B, then all those not beginning with A&gt;B&gt;C: quickly the set of sequences to be tested diminishes.\u00a0<\/li>\n<\/ul>\n<p>These are the broad outlines of the algorithm, but you can find more detailed information in the\u00a0<a href=\"http:\/\/hanj.cs.illinois.edu\/pdf\/span01.pdf\" target=\"_blank\" rel=\"noreferrer noopener\">original article<\/a>.\u00a0<\/p>\n<h2><strong>Summing up<\/strong>\u00a0<\/h2>\n<p>We have set ourselves an objective:\u00a0<strong>to identify informative sub-sequences on a conversion act.<\/strong>\u00a0<\/p>\n<p>We have quantified the notion of information subsequence using\u00a0<strong>mutual information.<\/strong>\u00a0<\/p>\n<p>We have a routine for generating candidate sequences, with the guarantee that they are frequent:\u00a0<strong>PrefixSpan<\/strong>. This step requires a minimum frequency of appearance and a possible maximum sequence length.\u00a0<\/p>\n<p>The process of extracting informative\u00a0sequences consists of generating candidate sequences, then accepting or rejecting them according to the criterion of mutual information, or to a lesser extent, ordering them according to the amount of information they provide.\u00a0<\/p>\n<h2><strong>Analytics benefits<\/strong>\u00a0<\/h2>\n<h3><strong>From a descriptive point of view<\/strong>\u00a0<\/h3>\n<p>This work makes it possible to develop an advanced sequence exploration product to complement the\u00a0<a href=\"https:\/\/www.atinternet.com\/en\/resources\/resources\/understand-visitor-paths-with-navigation\/\" target=\"_blank\" rel=\"noreferrer noopener\">navigation<\/a>\u00a0module. An analyst could filter the sequences according to a minimal presence and analyse the effect of such sequences on the conversion rate. Alternatively, they could keep only the sequences associated with the highest \/ lowest conversion rates and take actions to promote \/ avoid such sequences. Finally, by setting conversion and frequency thresholds, any AT Internet user could quickly identify which are the key sequences of their visitors.\u00a0<\/p>\n<h3><strong>From a predictive point of view<\/strong>\u00a0<\/h3>\n<p>As part of a project to identify hesitant shoppers, we began by developing a model for learning purchasing behaviour based on the simple metrics we described at the beginning of this article: page views, visit time, number of visits, etc. Then, as the model was iterated, we injected variables into it to indicate the presence of sequences that had been selected as informative using the procedure described above. The addition of these variables allowed us to improve our prediction metrics by 5 to 10% depending on the cases and they were therefore retained in our\u00a0modelling.\u00a0<\/p>\n<h3><strong>To go even further<\/strong>\u00a0<\/h3>\n<p>An informed reader will notice some shortcuts to this presentation. In order to avoid overburdening the technical aspects, a few problems have been left out, such as:\u00a0<\/p>\n<ul>\n<li>The imbalance of classes means that user-specific sequences that convert carry little weight in the calculation of\u00a0PrefixSpan\u00a0support. For example, if the conversion rate is only 1% and a discriminating sequence is only present in the buyer group, its frequency will only weigh for\u00a0<strong>at most<\/strong>\u00a01% in the calculation of the minimum support (which was more like 20% in our uses, for a quality\/quantity compromise of the sequences) and the sequence will be filtered. It is advisable to carry out the extraction treatment separately for the two groups (navigation sequences\u00a0with and without conversion).\u00a0\u00a0<\/li>\n<li>As sequences are extracted using a univariate criterion (only one sequence is considered at a time), it is\u00a0common for similar sequences or sequences bearing the same information to be extracted. Post-processing by looking for co-occurrences may be necessary to reduce information redundancy.\u00a0<\/li>\n<li>Finally, a\u00a0<a href=\"https:\/\/link.springer.com\/article\/10.1007\/s10115-014-0817-0\" target=\"_blank\" rel=\"noreferrer noopener\">paper<\/a>\u00a0co-authored by Amazon and Siemens suggests a sequence extraction procedure that takes on board the notion of selection during sequence generation (based on a different algorithm: BIDE), which makes it possible to prune the tree of possible sequences all the more efficiently. This approach seems\u00a0promising,\u00a0but it was not initially explored because of its implementation cost.\u00a0<\/li>\n<\/ul>\n<\/div>\n\n","protected":false},"excerpt":{"rendered":"<p>When it comes to analysing behavioural web analytics data, analysts often work with aggregates: visit duration, number of page views, bounce rate\u2026 These variables make it possible to describe each navigation sequence in a very structured way. They are easy to represent for analysis purposes, or to use in a machine learning environment for example.\u00a0 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":59391,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[12033],"tags":[10990,11603,33276,33275,22019],"dealstore":[],"offerexpiration":[],"class_list":["post-59390","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-analytics","tag-analysis","tag-data","tag-navigational","tag-refine","tag-sequential"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>How do you use sequential data to refine navigational analysis\u2753 - Som2ny Network<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/fivemor.com\/?p=59390\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"How do you use sequential data to refine navigational analysis\u2753 - Som2ny Network\" \/>\n<meta property=\"og:description\" content=\"When it comes to analysing behavioural web analytics data, analysts often work with aggregates: visit duration, number of page views, bounce rate\u2026 These variables make it possible to describe each navigation sequence in a very structured way. They are easy to represent for analysis purposes, or to use in a machine learning environment for example.\u00a0 [&hellip;]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/fivemor.com\/?p=59390\" \/>\n<meta property=\"og:site_name\" content=\"Som2ny Network\" \/>\n<meta property=\"article:published_time\" content=\"2025-01-31T06:31:03+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1000\" \/>\n\t<meta property=\"og:image:height\" content=\"750\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"10 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/fivemor.com\/?p=59390#article\",\"isPartOf\":{\"@id\":\"https:\/\/fivemor.com\/?p=59390\"},\"author\":{\"name\":\"admin\",\"@id\":\"https:\/\/fivemor.com\/#\/schema\/person\/b85e3c3dc0e1daea076524dc8810c371\"},\"headline\":\"How do you use sequential data to refine navigational analysis\u2753\",\"datePublished\":\"2025-01-31T06:31:03+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/fivemor.com\/?p=59390\"},\"wordCount\":1997,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\/\/fivemor.com\/#organization\"},\"image\":{\"@id\":\"https:\/\/fivemor.com\/?p=59390#primaryimage\"},\"thumbnailUrl\":\"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg\",\"keywords\":[\"Analysis\",\"Data\",\"navigational\",\"refine\",\"Sequential\"],\"articleSection\":[\"Analytics\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/fivemor.com\/?p=59390#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/fivemor.com\/?p=59390\",\"url\":\"https:\/\/fivemor.com\/?p=59390\",\"name\":\"How do you use sequential data to refine navigational analysis\u2753 - Som2ny Network\",\"isPartOf\":{\"@id\":\"https:\/\/fivemor.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/fivemor.com\/?p=59390#primaryimage\"},\"image\":{\"@id\":\"https:\/\/fivemor.com\/?p=59390#primaryimage\"},\"thumbnailUrl\":\"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg\",\"datePublished\":\"2025-01-31T06:31:03+00:00\",\"breadcrumb\":{\"@id\":\"https:\/\/fivemor.com\/?p=59390#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/fivemor.com\/?p=59390\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/fivemor.com\/?p=59390#primaryimage\",\"url\":\"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg\",\"contentUrl\":\"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg\",\"width\":1000,\"height\":750},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/fivemor.com\/?p=59390#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/fivemor.com\/?bp_activities=1\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"How do you use sequential data to refine navigational analysis\u2753\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/fivemor.com\/#website\",\"url\":\"https:\/\/fivemor.com\/\",\"name\":\"Som2ny Network\",\"description\":\"Daily Deals\",\"publisher\":{\"@id\":\"https:\/\/fivemor.com\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/fivemor.com\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/fivemor.com\/#organization\",\"name\":\"Som2ny Network\",\"url\":\"https:\/\/fivemor.com\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/fivemor.com\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/fivemor.com\/wp-content\/uploads\/2026\/07\/4a0953c4-logo-300x86-1.png\",\"contentUrl\":\"https:\/\/fivemor.com\/wp-content\/uploads\/2026\/07\/4a0953c4-logo-300x86-1.png\",\"width\":300,\"height\":86,\"caption\":\"Som2ny Network\"},\"image\":{\"@id\":\"https:\/\/fivemor.com\/#\/schema\/logo\/image\/\"}},{\"@type\":\"Person\",\"@id\":\"https:\/\/fivemor.com\/#\/schema\/person\/b85e3c3dc0e1daea076524dc8810c371\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/fivemor.com\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/729ae85bf62b9917e93538db2f2688ca?s=96&r=g&default=https%3A%2F%2Ffivemor.com%2Fwp-content%2Fplugins%2Fbuddypress-first-letter-avatar%2Fimages%2Fdefault%2F96%2Flatin_a.png\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/729ae85bf62b9917e93538db2f2688ca?s=96&r=g&default=https%3A%2F%2Ffivemor.com%2Fwp-content%2Fplugins%2Fbuddypress-first-letter-avatar%2Fimages%2Fdefault%2F96%2Flatin_a.png\",\"caption\":\"admin\"},\"sameAs\":[\"https:\/\/fivemor.com\"],\"url\":\"https:\/\/fivemor.com\/?author=1\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"How do you use sequential data to refine navigational analysis\u2753 - Som2ny Network","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/fivemor.com\/?p=59390","og_locale":"en_US","og_type":"article","og_title":"How do you use sequential data to refine navigational analysis\u2753 - Som2ny Network","og_description":"When it comes to analysing behavioural web analytics data, analysts often work with aggregates: visit duration, number of page views, bounce rate\u2026 These variables make it possible to describe each navigation sequence in a very structured way. They are easy to represent for analysis purposes, or to use in a machine learning environment for example.\u00a0 [&hellip;]","og_url":"https:\/\/fivemor.com\/?p=59390","og_site_name":"Som2ny Network","article_published_time":"2025-01-31T06:31:03+00:00","og_image":[{"width":1000,"height":750,"url":"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg","type":"image\/jpeg"}],"author":"admin","twitter_card":"summary_large_image","twitter_misc":{"Written by":"admin","Est. reading time":"10 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/fivemor.com\/?p=59390#article","isPartOf":{"@id":"https:\/\/fivemor.com\/?p=59390"},"author":{"name":"admin","@id":"https:\/\/fivemor.com\/#\/schema\/person\/b85e3c3dc0e1daea076524dc8810c371"},"headline":"How do you use sequential data to refine navigational analysis\u2753","datePublished":"2025-01-31T06:31:03+00:00","mainEntityOfPage":{"@id":"https:\/\/fivemor.com\/?p=59390"},"wordCount":1997,"commentCount":0,"publisher":{"@id":"https:\/\/fivemor.com\/#organization"},"image":{"@id":"https:\/\/fivemor.com\/?p=59390#primaryimage"},"thumbnailUrl":"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg","keywords":["Analysis","Data","navigational","refine","Sequential"],"articleSection":["Analytics"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/fivemor.com\/?p=59390#respond"]}]},{"@type":"WebPage","@id":"https:\/\/fivemor.com\/?p=59390","url":"https:\/\/fivemor.com\/?p=59390","name":"How do you use sequential data to refine navigational analysis\u2753 - Som2ny Network","isPartOf":{"@id":"https:\/\/fivemor.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/fivemor.com\/?p=59390#primaryimage"},"image":{"@id":"https:\/\/fivemor.com\/?p=59390#primaryimage"},"thumbnailUrl":"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg","datePublished":"2025-01-31T06:31:03+00:00","breadcrumb":{"@id":"https:\/\/fivemor.com\/?p=59390#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/fivemor.com\/?p=59390"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/fivemor.com\/?p=59390#primaryimage","url":"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg","contentUrl":"https:\/\/fivemor.com\/wp-content\/uploads\/2025\/01\/sequential-data-science-.jpg","width":1000,"height":750},{"@type":"BreadcrumbList","@id":"https:\/\/fivemor.com\/?p=59390#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/fivemor.com\/?bp_activities=1"},{"@type":"ListItem","position":2,"name":"How do you use sequential data to refine navigational analysis\u2753"}]},{"@type":"WebSite","@id":"https:\/\/fivemor.com\/#website","url":"https:\/\/fivemor.com\/","name":"Som2ny Network","description":"Daily Deals","publisher":{"@id":"https:\/\/fivemor.com\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/fivemor.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/fivemor.com\/#organization","name":"Som2ny Network","url":"https:\/\/fivemor.com\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/fivemor.com\/#\/schema\/logo\/image\/","url":"https:\/\/fivemor.com\/wp-content\/uploads\/2026\/07\/4a0953c4-logo-300x86-1.png","contentUrl":"https:\/\/fivemor.com\/wp-content\/uploads\/2026\/07\/4a0953c4-logo-300x86-1.png","width":300,"height":86,"caption":"Som2ny Network"},"image":{"@id":"https:\/\/fivemor.com\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/fivemor.com\/#\/schema\/person\/b85e3c3dc0e1daea076524dc8810c371","name":"admin","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/fivemor.com\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/729ae85bf62b9917e93538db2f2688ca?s=96&r=g&default=https%3A%2F%2Ffivemor.com%2Fwp-content%2Fplugins%2Fbuddypress-first-letter-avatar%2Fimages%2Fdefault%2F96%2Flatin_a.png","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/729ae85bf62b9917e93538db2f2688ca?s=96&r=g&default=https%3A%2F%2Ffivemor.com%2Fwp-content%2Fplugins%2Fbuddypress-first-letter-avatar%2Fimages%2Fdefault%2F96%2Flatin_a.png","caption":"admin"},"sameAs":["https:\/\/fivemor.com"],"url":"https:\/\/fivemor.com\/?author=1"}]}},"_links":{"self":[{"href":"https:\/\/fivemor.com\/index.php?rest_route=\/wp\/v2\/posts\/59390","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/fivemor.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/fivemor.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/fivemor.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/fivemor.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=59390"}],"version-history":[{"count":0,"href":"https:\/\/fivemor.com\/index.php?rest_route=\/wp\/v2\/posts\/59390\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/fivemor.com\/index.php?rest_route=\/wp\/v2\/media\/59391"}],"wp:attachment":[{"href":"https:\/\/fivemor.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=59390"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/fivemor.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=59390"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/fivemor.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=59390"},{"taxonomy":"dealstore","embeddable":true,"href":"https:\/\/fivemor.com\/index.php?rest_route=%2Fwp%2Fv2%2Fdealstore&post=59390"},{"taxonomy":"offerexpiration","embeddable":true,"href":"https:\/\/fivemor.com\/index.php?rest_route=%2Fwp%2Fv2%2Fofferexpiration&post=59390"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}