I\u2019ve been writing about data and analytics for a while now, and there are some terms and phrases that I\u2019ve used so often that I\u2019m sure I know what they mean. I can generally use them in the right context, but I would be hard-pressed to come up with a definition.<\/p>\n
![\"White](\"https:\/\/www.dataiq.global\/devstage\/wp-content\/uploads\/domino-3192542_1280.jpg\" "\\"\\"")
Thankfully Data Mettle, a \u201cdata science workshop,\u201d held a presentation and Q&A session to explain neural networks and other data science basics. Data Mettle founder, Jeremy Mitchel,l said that neural networks are a type of machine learning model and then displayed an image of a neuron.<\/p>\n
\u201cIt receives signals into these inputs and if those signals are big enough or strong enough, it will shoot an electrical pulse down the length of the nerve to the other side,\u201d he said, adding the disclaimer that a neurologist might not explain the process in the same way.<\/p>\n
He then showed us something similar to a cat\u2019s cradle diagram with several dots connected by lines. This represented a series of neurons. The signal would pass from left to right. I imagined it as being similar to a domino run.<\/p>\n
Mitchell explained that this type of network could be used by data scientists for image recognition. One use case could be a furniture retailer, wanting to classify the images of its stock. The image would be the input while the classification of a bed, chair, table, etc, would be the output.<\/p>\n
\nThe network had to be trained with thousands of examples.<\/h3>\n<\/blockquote>\n
\u201cWe put our input signal in the left and those signals get passed down and down and down and then, hopefully, what we get at the end is one or maybe two things lit up. We say \u2018OK. We put this image in and this one\u2019s lit up, so therefore it is a table\u2019,\u201d said Mitchell. I got a bit lost. I wasn\u2019t sure if the path along the connections represented something like a digital fingerprint and therefore input that generates similar-shaped paths would have the same output.<\/p>\n
That aside, the rest was fairly understandable. The network had to be trained with thousands of examples of furniture images. \u201cYou get all your example data and you use that to infer the model. Once you\u2019ve got that, you can take an image you\u2019ve never seen before, pass it in on the left and then you get your classification,\u201d stated Mitchell.<\/p>\n
He said this is the type of technology used in facial recognition in photos. It is also used by YouTube to classify the videos uploaded to its platform, and by Facebook to help identify inappropriate content.<\/p>\n
Mitchell then stated that, while a lot of focus on is on deep learning and neural networks, there are plenty of tools to use in the data science space, such as recommender engines and predictors. Data Mettle worked with a client to predict the size of investment tech start-ups would receive.<\/p>\n
\u201cYou need a lot of examples, thousands of examples of companies seeking investment,\u201d said Mitchell. By running analytics on those example, the Data Mettle team could predict what characteristics companies needed to have in order to land investment of a certain size. Once they created that formula, different start-ups with different characteristics could be fed into the calculation, resulting in a prediction of the amount of funding they would be able to secure.<\/p>\n
Now I understand that recommender engines and predictors are simply mathematical formulas. At least that wasn\u2019t so hard to grasp.<\/p>\n","protected":false},"excerpt":{"rendered":"
I\u2019ve been writing about data and analytics for a while now, and there are some terms and phrases that I\u2019ve used so often that I\u2019m sure I know what they mean. I can generally use them in the right context, but I would be hard-pressed to come up wit…<\/p>\n","protected":false},"author":3,"featured_media":15488,"menu_order":0,"comment_status":"open","ping_status":"closed","template":"","format":"standard","meta":{"_acf_changed":false,"content-type":"","_searchwp_excluded":"","footnotes":""},"categories":[129,398],"tags":[169,175,121,91],"pillar":[],"class_list":["post-15485","article","type-article","status-publish","format-standard","has-post-thumbnail","hentry","category-editorial","category-public","tag-analytics-and-insight","tag-data-science","tag-skills","tag-technology"],"acf":[],"publishpress_future_action":{"enabled":false,"date":"2026-05-21 07:45:08","action":"change-status","newStatus":"draft","terms":[],"taxonomy":"category","extraData":[]},"publishpress_future_workflow_manual_trigger":{"enabledWorkflows":[]},"_links":{"self":[{"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/article\/15485","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/article"}],"about":[{"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/types\/article"}],"author":[{"embeddable":true,"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/comments?post=15485"}],"version-history":[{"count":0,"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/article\/15485\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/media\/15488"}],"wp:attachment":[{"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/media?parent=15485"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/categories?post=15485"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/tags?post=15485"},{"taxonomy":"pillar","embeddable":true,"href":"https:\/\/www.dataiq.global\/devstage\/wp-json\/wp\/v2\/pillar?post=15485"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}