<?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 Denis Semenenko on Medium]]></title>
        <description><![CDATA[Stories by Denis Semenenko on Medium]]></description>
        <link>https://medium.com/@denny_sem?source=rss-7adc29e2905------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*3sidwLXa4zN5L8TQl0Qn5Q.jpeg</url>
            <title>Stories by Denis Semenenko on Medium</title>
            <link>https://medium.com/@denny_sem?source=rss-7adc29e2905------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 13:03:51 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@denny_sem/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[Data Scientist Resume Projects]]></title>
            <link>https://medium.com/cube-dev/data-scientist-resume-projects-806a74388ae6?source=rss-7adc29e2905------2</link>
            <guid isPermaLink="false">https://medium.com/p/806a74388ae6</guid>
            <category><![CDATA[resume]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[towards-data-science]]></category>
            <category><![CDATA[data-scientist]]></category>
            <dc:creator><![CDATA[Denis Semenenko]]></dc:creator>
            <pubDate>Thu, 22 Jun 2017 12:40:17 GMT</pubDate>
            <atom:updated>2019-03-07T14:56:32.890Z</atom:updated>
            <content:encoded><![CDATA[<h4>Machine learning problems set to build a data scientist CV without work experience</h4><figure><a href="https://cube.dev/"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jdquSr0C9xKLn4vQPC3fZQ.png" /></a></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3qUMeSLz7IR0pyLa4KUgXA.jpeg" /></figure><p>Data scientists are one of the most hirable specialists today, but it’s not so easy to enter this profession without a “Projects” field in your resume. You need experience to get the job, and you need the job to get the experience. Seems like a vicious circle, right?</p><p><a href="https://statsbot.co/product/overview?utm_source=blog&amp;utm_medium=article&amp;utm_campaign=resume_projects">Statsbot</a>’s data scientist Denis Semenenko wrote this article to help everyone with making the first simple, but yet illustrative data science projects which can take less than a week of work time.</p><blockquote>The great advantage of these projects is that each of them is a full-stack data science problem.</blockquote><p>This means that you need to formulate the problem, design the solution, find the data, master the technology, build a machine learning model, evaluate the quality, and maybe wrap it into a simple UI. This is a more diverse approach than, for example, Kaggle competition or Coursera lessons (but they are quite good too!).</p><p>Keep reading if you want to improve your CV by using a data science project, find ideas for a university project, or just practice in a particular domain of machine learning.</p><h3><strong>Spam or Ham</strong></h3><p>Spam lives wherever it’s possible to leave messages. One of the classic data science problems is a spam detection. You can train a model for detecting spam emails, spam messages, and spam user comments to hide them in browser.</p><blockquote>Look at <a href="https://blog.statsbot.co/text-classifier-algorithms-in-machine-learning-acc115293278"><strong>the key text classifier algorithms with use cases and tutorials</strong></a>.</blockquote><p>A machine learning engine defines spam based on the probability of meeting such words as “sale” and “buy” in spam messages. As a result, you can get a working prototype of AdBlock in about one week.</p><p><strong>ML problem:</strong> <a href="https://en.wikipedia.org/wiki/Document_classification">text classification</a><br><strong>Algorithms:</strong> naive bayes, linear classifiers, tree classifiers, whatever-you-want classifiers<br><strong>Technologies:</strong> <a href="http://scikit-learn.org/stable/">sklearn</a>, <a href="http://www.nltk.org/">nltk</a>, <a href="https://scrapy.org/">scrapy</a><br><strong>Data:</strong> <a href="https://www.kaggle.com/uciml/sms-spam-collection-dataset">sms spam dataset</a>, <a href="http://csmining.org/index.php/spam-email-datasets-.html">e-mail spam dataset</a> , <a href="https://archive.ics.uci.edu/ml/datasets/YouTube+Spam+Collection">youtube comments spam dataset</a><br><strong>Implementation:</strong> browser extension<br><strong>References: </strong><a href="https://adblockplus.org/">AdBlock</a>, <a href="http://adguard.com/">Adguard</a><br><strong>Guides:</strong> <a href="https://hackernoon.com/how-to-build-a-simple-spam-detecting-machine-learning-classifier-4471fe6b816e">How To Build a Simple Spam-Detecting Machine Learning Classifier</a>, <a href="https://developer.chrome.com/extensions/getstarted">Getting Started: Building a Chrome Extension</a></p><h3><strong>Not Hotdog</strong></h3><p>Not Hotdog is a viral app from the Silicon Valley series which can recognize hotdogs and non-hotdogs in photos.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FACmydtFDTGs%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DACmydtFDTGs&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FACmydtFDTGs%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/a3f32092a676b2dbce28256ae03a6cd6/href">https://medium.com/media/a3f32092a676b2dbce28256ae03a6cd6/href</a></iframe><p>You can train an image classifier for whatever you want by using social networks or google images to collect a dataset (using hashtags or search queries).</p><p>Training of a neural network from scratch requires a lot of train samples and computing time, so it is better to use a pre-trained network (this approach is called transfer learning).</p><p><strong>ML problem:</strong> <a href="https://en.wikipedia.org/wiki/Computer_vision#Recognition">image recognition</a>, <a href="https://en.wikipedia.org/wiki/Contextual_image_classification">image classification</a>, <a href="https://en.wikipedia.org/wiki/Transfer_learning">transfer learning</a><br><strong>Algorithms:</strong> convolutional neural networks<br><strong>Technologies:</strong> <a href="https://keras.io/">keras</a>, <a href="https://lasagne.readthedocs.io/en/latest/">lasagne</a>, <a href="https://www.instagram.com/developer/">Instagram API</a>(or external libraries e.g.<a href="https://github.com/instagrambot"> Instabot</a>)<br><strong>Data:</strong> use Instagram API and hashtags to collect dataset<br><strong>Implementation:</strong> mobile app<br><strong>References:</strong> <a href="https://itunes.apple.com/us/app/not-hotdog/id1212457521?mt=8">Not hotdog</a><br><strong>Guides:</strong> <a href="https://medium.com/towards-data-science/transfer-learning-using-keras-d804b2e04ef8">Transfer learning using Keras</a>, <a href="https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html">Building powerful image classification models using very little data</a></p><h3><strong>Netflix movie recommendations</strong></h3><p>Recommender systems are necessary for large companies like Google or Facebook, because it is valuable from a perspective of revenue and engagement (Facebook ads, Youtube recommendations).</p><p>Data scientist beginners are able to get practice in this sub-domain of data science and build their personal movie recommender system.</p><blockquote>For those who want to go deeper with this approach, the <a href="https://statsbot.co/product/predictions?utm_source=blog&amp;utm_medium=article&amp;utm_campaign=resume_projects">Statsbot</a>’s team has prepared the <a href="https://blog.statsbot.co/recommendation-system-algorithms-ba67f39ac9a3"><strong>overview of the main recommendation system algorithms</strong></a>.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*wuPyJMvP5QBSbhC3." /><figcaption>Illustration: movielens.org</figcaption></figure><p><strong>ML problem:</strong> <a href="https://en.wikipedia.org/wiki/Recommender_system">recommender systems</a><br><strong>Algorithms:</strong> dimensionality reduction, collaborative filtering, classification algorithms<br><strong>Technologies:</strong> <a href="http://scikit-learn.org/stable/">sklearn</a>, <a href="https://github.com/JohnLangford/vowpal_wabbit/wiki">vowpal wabbit</a><br><strong>Data:</strong> <a href="http://academictorrents.com/details/9b13183dc4d60676b773c9e2cd6de5e5542cee9a">Netflix prize dataset</a>, <a href="https://grouplens.org/datasets/movielens/">MovieLens dataset</a><br><strong>Implementation:</strong> web application<br><strong>References:</strong> <a href="http://jinni.com/">Jinni</a>, <a href="http://movielens.org/">MovieLens</a><br><strong>Guides:</strong> <a href="https://www.analyticsvidhya.com/blog/2016/06/quick-guide-build-recommendation-engine-python/">Quick Guide to Build a Recommendation Engine in Python</a></p><h3><strong>Original Snapchat lenses</strong></h3><p>If you are good at graphical design you should try to create your own custom Snapchat lens. Any social network has such things in their arsenal — Instagram, Facebook, and Snapchat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/375/0*gL5rm-tzYdDcHgag." /><figcaption>Gif: <a href="https://support.snapchat.com/en-GB/a/lenses1">https://support.snapchat.com/en-GB/a/lenses1</a></figcaption></figure><p>Lenses detect key facial points to show where the borders of the lips, eyes, nose, and face are. Then, ML engine builds a mask matching the face correctly even if you’re moving (for example, using openCV or any graphic library).</p><p><strong>ML problem:</strong> <a href="https://en.wikipedia.org/wiki/Computer_vision#Recognition">image recognition</a>,<a href="https://en.wikipedia.org/wiki/Face_detection"> face detection</a><br><strong>Algorithms:</strong> convolutional neural networks, facial keypoints detection<br><strong>Technologies:</strong> <a href="http://dlib.net/">dlib</a>, <a href="https://github.com/cmusatyalab/openface">openface</a>, <a href="https://keras.io/">keras</a>, <a href="http://opencv.org/">openCV</a><br><strong>Data:</strong><a href="https://www.kaggle.com/c/facial-keypoints-detection"> Facial keypoints detection dataset</a><br><strong>Implementation:</strong> mobile app<br><strong>References:</strong> <a href="https://www.snapchat.com/">Snapchat</a>, <a href="http://www.instagram.com/">Instagram</a><br><strong>Guides:</strong> <a href="http://www.pyimagesearch.com/2017/04/03/facial-landmarks-dlib-opencv-python/">Facial landmarks with dlib, OpenCV, and Python</a>, <a href="https://www.appcoda.com/ios-programming-camera-iphone-app/">Build a Simple Camera App</a></p><h3><strong>Twitter stream</strong></h3><p>A lot of companies monitor mentions from their customers on the internet to react to the negative ones ASAP. For example, T-Mobile and Verizon need to respond to negative tweets fast and find out what the problem is and how they can solve it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ymruPxDm89zkJhef." /></figure><p>Everyone can make this project using a convenient Twitter API and sentiment analysis algorithms to detect such tweets in the whole stream.</p><p><strong>ML problem:</strong> <a href="https://en.wikipedia.org/wiki/Sentiment_analysis">sentiment analysis</a><br><strong>Algorithms:</strong> sentiment analysis<br><strong>Data:</strong> <a href="https://dev.twitter.com/rest/public">Twitter API</a><br><strong>Technologies:</strong> <a href="http://www.nltk.org/">nltk</a>, <a href="https://spacy.io/">spaCy</a><br><strong>Implementation: </strong>web app<br><strong>References:</strong> <a href="http://twilert.com/">Twilert</a>, <a href="http://tweetreach.com/">Tweetreach</a><br><strong>Guides:</strong> <a href="http://www.laurentluce.com/posts/twitter-sentiment-analysis-using-python-and-nltk/">Twitter sentiment analysis using Python and NLTK</a></p><h3><strong>Tennis betmaker</strong></h3><p>Professional betmakers always look for profitable bets. Tennis is a good kind of sport for predictions due to large amounts of statistical data, and sometimes bookmakers make mistakes.</p><p>Data scientists can put together websites with match history and players’ information, build a prediction model, and compare the results with bookmaker’s evaluations.</p><p>The goals is to find a gap between ML prediction and a bookmaker’s evaluation, which gives a chance to win. It’s a good data scientist problem for feature engineering!</p><p><strong>ML problem:</strong> <a href="https://en.wikipedia.org/wiki/Classification">classification</a><br><strong>Algorithms:</strong> classification<br><strong>Data: </strong>atpworldtour.com<br><strong>Technologies:</strong> <a href="http://scikit-learn.org/stable/">sklearn</a>, <a href="https://scrapy.org/">scrapy</a><br><strong>Implementation:</strong> web app<br><strong>References: </strong>olbg.com, verifiedbets.com<br><strong>Guides:</strong> <a href="http://www.doc.ic.ac.uk/teaching/distinguished-projects/2015/m.sipko.pdf">Machine Learning for the Prediction of Professional Tennis Matches</a></p><h3><strong>Stock price prediction</strong></h3><p>A popular application of ML is time series prediction. A data science engine can predict exchange rates and stocks, so traders or bots can gamble based on these predictions.</p><p>If you choose this problem, you’ll find out that it’s easy to get such data and practice on it. This sub-domain is derived from econometrics and classic machine learning, so you should be ready to explore a statistics method.</p><p><strong>ML problem:</strong> <a href="https://en.wikipedia.org/wiki/Time_series#Prediction_and_forecasting">time series prediction</a><br><strong>Algorithms:</strong> ARIMA, regression<br><strong>Technologies:</strong> <a href="http://scikit-learn.org/stable/">sklearn</a>, <a href="https://github.com/facebookincubator/prophet">prophet</a>, <a href="https://scrapy.org/">scrapy</a><br><strong>Data:</strong> <a href="https://www.quandl.com/search?query=">Quandl</a><br><strong>Implementation:</strong> web app<br><strong>References:</strong> financeboards.com<br><strong>Guides:</strong> <a href="https://ntguardian.wordpress.com/2016/09/19/introduction-stock-market-data-python-1/">An Introduction to Stock Market Data Analysis with Python</a></p><p>I hope you enjoy these toy problems in data science! Using them, it’s simple to improve your data scientist resume and get a high-demand job.</p><p>If you know some more relevant ML projects, please share them in the comments below and I will include them in an article. And of course, recommend this post if you find it helpful for you or other data science lovers.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F0d3e8c%3Fas_embed%3Dtrue&amp;url=https%3A%2F%2Fupscri.be%2F0d3e8c%2F&amp;image=https%3A%2F%2Fupscri.be%2Fmedia%2Fform.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/02231cd5403151a2a463e36cc3b88462/href">https://medium.com/media/02231cd5403151a2a463e36cc3b88462/href</a></iframe><h3>YOU’D ALSO LIKE:</h3><ul><li><a href="https://blog.statsbot.co/machine-learning-devops-611210393c1a">DevOps Pipeline for a Machine Learning Project</a></li><li><a href="https://blog.statsbot.co/machine-learning-translation-96f0ed8f19e4">Machine Learning Translation and the Google Translate Algorithm</a></li><li><a href="https://blog.statsbot.co/recommendation-system-algorithms-ba67f39ac9a3">Recommendation System Algorithms</a></li><li><a href="https://blog.statsbot.co/text-classifier-algorithms-in-machine-learning-acc115293278">Text Classifier Algorithms in Machine Learning</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=806a74388ae6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/cube-dev/data-scientist-resume-projects-806a74388ae6">Data Scientist Resume Projects</a> was originally published in <a href="https://medium.com/cube-dev">Cube Dev</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>