ºÝºÝߣshows by User: kitsamho / http://www.slideshare.net/images/logo.gif ºÝºÝߣshows by User: kitsamho / Mon, 04 Feb 2019 12:54:20 GMT ºÝºÝߣShare feed for ºÝºÝߣshows by User: kitsamho Classifying brands on Facebook using supervised machine learning /slideshow/classifying-brands-on-facebook-using-supervised-machine-learning/130472916 capstonedeck-190204125420
Context: Facebook pages are a significant brand asset for thousands of companies and organisations worldwide. Companies invest heavily into developing social content to engage with customers and prospects in a two way 'conversation’. The impact of social media on business success is widely debated (often with very differing views) however one thing that most marketers agree on is that brand differentiation is a key aspect of any 'healthy' brand. Brands need to stand out from one another in their category. Not just in terms of what they offer but also in terms of how they communicate - across all media channels, including social media. Which leads to the focus of this project: Are brands doing enough to differentiate their social content on Facebook? Goal: Focussing on the seven biggest UK supermarket brands on Facebook and using natural language processing and supervised classification modelling, can we train a machine to distinguish the different supermarket brands' social content from one another on Facebook? Approach: Seven of the UK's leading supermarket brands were chosen for the study: Sainsbury's, Tesco, Lidl, ASDA, Morrisons, M&S and Waitrose. Their social content was scraped from Facebook with automated web scraping (Selenium). A total of 6350 posts were scraped from c.2014 to early December 2019, and after cleaning and debranding was performed we managed to have the following distribution of brands, with a baseline of 0.21 Results: After applying term frequency-inverse document frequency (TF-IDF) vectorization and then implementing a tuned framework of supervised classifiers (Logistic Regression, RandomForest, KNN, SVM and a Multi-Layer Perceptron) to model the Facebook content, I managed to achieve an accuracy score of 0.70 using an ensemble voting classifier, which exceeded my baseline of 0.21. My results indicate that supermarket brands in the UK are distinct in terms of their Facebook content with the key differentiating features being broadly aligned with the brands’ overarching strategies we see across all their other marcomm channels.]]>
Context: Facebook pages are a significant brand asset for thousands of companies and organisations worldwide. Companies invest heavily into developing social content to engage with customers and prospects in a two way 'conversation’. The impact of social media on business success is widely debated (often with very differing views) however one thing that most marketers agree on is that brand differentiation is a key aspect of any 'healthy' brand. Brands need to stand out from one another in their category. Not just in terms of what they offer but also in terms of how they communicate - across all media channels, including social media. Which leads to the focus of this project: Are brands doing enough to differentiate their social content on Facebook? Goal: Focussing on the seven biggest UK supermarket brands on Facebook and using natural language processing and supervised classification modelling, can we train a machine to distinguish the different supermarket brands' social content from one another on Facebook? Approach: Seven of the UK's leading supermarket brands were chosen for the study: Sainsbury's, Tesco, Lidl, ASDA, Morrisons, M&S and Waitrose. Their social content was scraped from Facebook with automated web scraping (Selenium). A total of 6350 posts were scraped from c.2014 to early December 2019, and after cleaning and debranding was performed we managed to have the following distribution of brands, with a baseline of 0.21 Results: After applying term frequency-inverse document frequency (TF-IDF) vectorization and then implementing a tuned framework of supervised classifiers (Logistic Regression, RandomForest, KNN, SVM and a Multi-Layer Perceptron) to model the Facebook content, I managed to achieve an accuracy score of 0.70 using an ensemble voting classifier, which exceeded my baseline of 0.21. My results indicate that supermarket brands in the UK are distinct in terms of their Facebook content with the key differentiating features being broadly aligned with the brands’ overarching strategies we see across all their other marcomm channels.]]> Mon, 04 Feb 2019 12:54:20 GMT /slideshow/classifying-brands-on-facebook-using-supervised-machine-learning/130472916 kitsamho@slideshare.net(kitsamho) Classifying brands on Facebook using supervised machine learning kitsamho Context: Facebook pages are a significant brand asset for thousands of companies and organisations worldwide. Companies invest heavily into developing social content to engage with customers and prospects in a two way 'conversation’. The impact of social media on business success is widely debated (often with very differing views) however one thing that most marketers agree on is that brand differentiation is a key aspect of any 'healthy' brand. Brands need to stand out from one another in their category. Not just in terms of what they offer but also in terms of how they communicate - across all media channels, including social media. Which leads to the focus of this project: Are brands doing enough to differentiate their social content on Facebook? Goal: Focussing on the seven biggest UK supermarket brands on Facebook and using natural language processing and supervised classification modelling, can we train a machine to distinguish the different supermarket brands' social content from one another on Facebook? Approach: Seven of the UK's leading supermarket brands were chosen for the study: Sainsbury's, Tesco, Lidl, ASDA, Morrisons, M&S and Waitrose. Their social content was scraped from Facebook with automated web scraping (Selenium). A total of 6350 posts were scraped from c.2014 to early December 2019, and after cleaning and debranding was performed we managed to have the following distribution of brands, with a baseline of 0.21 Results: After applying term frequency-inverse document frequency (TF-IDF) vectorization and then implementing a tuned framework of supervised classifiers (Logistic Regression, RandomForest, KNN, SVM and a Multi-Layer Perceptron) to model the Facebook content, I managed to achieve an accuracy score of 0.70 using an ensemble voting classifier, which exceeded my baseline of 0.21. My results indicate that supermarket brands in the UK are distinct in terms of their Facebook content with the key differentiating features being broadly aligned with the brands’ overarching strategies we see across all their other marcomm channels. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/capstonedeck-190204125420-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Context: Facebook pages are a significant brand asset for thousands of companies and organisations worldwide. Companies invest heavily into developing social content to engage with customers and prospects in a two way &#39;conversation’. The impact of social media on business success is widely debated (often with very differing views) however one thing that most marketers agree on is that brand differentiation is a key aspect of any &#39;healthy&#39; brand. Brands need to stand out from one another in their category. Not just in terms of what they offer but also in terms of how they communicate - across all media channels, including social media. Which leads to the focus of this project: Are brands doing enough to differentiate their social content on Facebook? Goal: Focussing on the seven biggest UK supermarket brands on Facebook and using natural language processing and supervised classification modelling, can we train a machine to distinguish the different supermarket brands&#39; social content from one another on Facebook? Approach: Seven of the UK&#39;s leading supermarket brands were chosen for the study: Sainsbury&#39;s, Tesco, Lidl, ASDA, Morrisons, M&amp;S and Waitrose. Their social content was scraped from Facebook with automated web scraping (Selenium). A total of 6350 posts were scraped from c.2014 to early December 2019, and after cleaning and debranding was performed we managed to have the following distribution of brands, with a baseline of 0.21 Results: After applying term frequency-inverse document frequency (TF-IDF) vectorization and then implementing a tuned framework of supervised classifiers (Logistic Regression, RandomForest, KNN, SVM and a Multi-Layer Perceptron) to model the Facebook content, I managed to achieve an accuracy score of 0.70 using an ensemble voting classifier, which exceeded my baseline of 0.21. My results indicate that supermarket brands in the UK are distinct in terms of their Facebook content with the key differentiating features being broadly aligned with the brands’ overarching strategies we see across all their other marcomm channels.
Classifying brands on Facebook using supervised machine learning from Sam Ho
]]> 117 1 https://cdn.slidesharecdn.com/ss_thumbnails/capstonedeck-190204125420-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Be kind, please rewind /slideshow/be-kind-please-rewind/8101651 samhobekindpleaserewind-110525151237-phpapp02
]]>
]]> Wed, 25 May 2011 15:12:34 GMT /slideshow/be-kind-please-rewind/8101651 kitsamho@slideshare.net(kitsamho) Be kind, please rewind kitsamho <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/samhobekindpleaserewind-110525151237-phpapp02-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br>
Be kind, please rewind from Sam Ho
]]> 395 3 https://cdn.slidesharecdn.com/ss_thumbnails/samhobekindpleaserewind-110525151237-phpapp02-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Dance music marketing done well - Anjunabeats /slideshow/lessons-learned01anjunabeats/4674114 anjunabeatscutdownfinalnoyt-100703180153-phpapp01
Dance music marketing is a funny old thing. Some labels do it well, although more often than not - they don't. Anjunabeats is an example of the former. Enjoy.]]>
Dance music marketing is a funny old thing. Some labels do it well, although more often than not - they don't. Anjunabeats is an example of the former. Enjoy.]]> Sat, 03 Jul 2010 18:01:47 GMT /slideshow/lessons-learned01anjunabeats/4674114 kitsamho@slideshare.net(kitsamho) Dance music marketing done well - Anjunabeats kitsamho Dance music marketing is a funny old thing. Some labels do it well, although more often than not - they don't. Anjunabeats is an example of the former. Enjoy. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/anjunabeatscutdownfinalnoyt-100703180153-phpapp01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Dance music marketing is a funny old thing. Some labels do it well, although more often than not - they don&#39;t. Anjunabeats is an example of the former. Enjoy.
Dance music marketing done well - Anjunabeats from Sam Ho
]]> 696 2 https://cdn.slidesharecdn.com/ss_thumbnails/anjunabeatscutdownfinalnoyt-100703180153-phpapp01-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 https://cdn.slidesharecdn.com/profile-photo-kitsamho-48x48.jpg?cb=1666077495 I am an analytically oriented insights professional with ten years experience at some of the best research and creative agencies in the UK, helping my clients bridge the path between data and insight. By combining this experience with new approaches in Data Science, I’m keen to take things to the next level both professionally and personally. https://cdn.slidesharecdn.com/ss_thumbnails/capstonedeck-190204125420-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/classifying-brands-on-facebook-using-supervised-machine-learning/130472916 Classifying brands on ... https://cdn.slidesharecdn.com/ss_thumbnails/samhobekindpleaserewind-110525151237-phpapp02-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/be-kind-please-rewind/8101651 Be kind, please rewind https://cdn.slidesharecdn.com/ss_thumbnails/anjunabeatscutdownfinalnoyt-100703180153-phpapp01-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/lessons-learned01anjunabeats/4674114 Dance music marketing ...