<?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 Michael Malatesta on Medium]]></title>
        <description><![CDATA[Stories by Michael Malatesta on Medium]]></description>
        <link>https://medium.com/@malatesta.mike?source=rss-2b4f0cda7280------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*LsvUotineXLQOEyqA3e0_g.jpeg</url>
            <title>Stories by Michael Malatesta on Medium</title>
            <link>https://medium.com/@malatesta.mike?source=rss-2b4f0cda7280------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 03:41:56 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@malatesta.mike/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[Object-Oriented Data Science]]></title>
            <link>https://medium.com/@malatesta.mike/whoever-said-data-scientist-dont-need-to-use-an-object-oriented-programming-framework-must-have-3ba47a019a18?source=rss-2b4f0cda7280------2</link>
            <guid isPermaLink="false">https://medium.com/p/3ba47a019a18</guid>
            <category><![CDATA[object-oriented]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Michael Malatesta]]></dc:creator>
            <pubDate>Sat, 09 Dec 2017 10:16:42 GMT</pubDate>
            <atom:updated>2017-12-09T10:37:43.713Z</atom:updated>
            <content:encoded><![CDATA[<p>Whoever said data scientist don’t need to use an object-oriented programming framework must have had a screw loose. With the abstraction OOP provides for many of the machine learning processes you will find yourself truely living the DRY (Don’t Repeat Yourself) life.</p><p>I recently set out to build a tool that I could use for a diverse range of machine learning projects that implements the common processes I typically use. I was amazed with what I was able to accomplish in just a short time. I’ve implemented common features like train/test splitting and dummifying but also more complex features like predictive null-value imputing and feature selection.</p><p>Now when I start a new project I just import my module and create a class that inherits from the desired class in my module. So far I only have a Predictor class but I plan on adding a Cluster class. To make things even easier I followed the Python Package Tutorial and made my package installable and upgradable with pip, which took a whopping 5 minutes. It won’t be available publicly just yet but hopefully soon. If you want to help out just fork from my github repo mmalates/ds_utils and submit a pull request! I’m also working on a sample project using ds_utils in my ‘tractors’ repo so check it out!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3ba47a019a18" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tableau on Linux]]></title>
            <link>https://medium.com/@malatesta.mike/tableau-on-linux-11f5837eb070?source=rss-2b4f0cda7280------2</link>
            <guid isPermaLink="false">https://medium.com/p/11f5837eb070</guid>
            <category><![CDATA[tableau]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[virtualbox]]></category>
            <dc:creator><![CDATA[Michael Malatesta]]></dc:creator>
            <pubDate>Mon, 20 Nov 2017 19:33:06 GMT</pubDate>
            <atom:updated>2017-11-20T19:33:06.617Z</atom:updated>
            <content:encoded><![CDATA[<p>I found Tableau to be a really clean UI for doing EDA on a dataset very quickly. The sad thing is that it isn’t available for install on any Linux distros. Running on Wine seemed like a poor option because it has only been tested on older versions of Tableau. So I invested about 5 minutes getting a Windows 10 ISO running on VirtualBox and I’m very satisfied. The Public version of Tableau lacks the ability to connect with the wide variety of data storage tools we use in Data Science but I think having connections to local files and Google Drive is fine for those just getting started with the application.</p><p>I also decided to get Tableau Server for Linux Beta running on my machine but the documentation for that is really lacking. After first relying on the error messages to get it going, which were helpful but not enough, I found a pretty simple explanation of how to get it going on AWS EC2 that worked for my machine. I’m aware that it’s not really practical to run a Tableau Server on my local machine but it was good practice and experience getting it running.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=11f5837eb070" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Flask Applications for Data Science with AWS ElasticBeanstalk]]></title>
            <link>https://medium.com/@malatesta.mike/flask-applications-for-data-science-with-aws-elasticbeanstalk-ad5751859fb0?source=rss-2b4f0cda7280------2</link>
            <guid isPermaLink="false">https://medium.com/p/ad5751859fb0</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[elastic-beanstalk]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[flask]]></category>
            <dc:creator><![CDATA[Michael Malatesta]]></dc:creator>
            <pubDate>Wed, 25 Oct 2017 17:57:29 GMT</pubDate>
            <atom:updated>2017-10-26T00:33:08.064Z</atom:updated>
            <content:encoded><![CDATA[<p>Using ElasticBeanstalk (EB) is a great way to deploy a Flask application. If you want your app to interactively fit models with SciPy/sklearn, work with DataFrames using Pandas, etc. you will need to install the packages required using pip on the EC2 instance associated with your EB environment. Simply including these packages in the ‘requirements.txt’ file will cause a failed deployment because ‘pip’ requires sudo permissions to install them. I found a solution that worked for me and some other colleagues and I’m hoping it will save you a headache.</p><p>Step 1: Create your application with a t2.medium EC2 instance, but remove packages like Pandas, Scipy, and sklearn from the ‘requirements.txt’ file.($ eb create -i t2.medium)(AWS has good instructions for deploying on EB) With less than 4GB of RAM Pandas and SciPy installs fail with a Memory Error. The app should deploy successfully but will fail when these packages are imported in your code. I create the environment interactively and choose Ohio (us-east-2). Make sure you set up ssh with a keypair.</p><p>Step 2: Use ssh to connect to the EC2 instance associated with your EB environment. (AWS has good documentation on how to do this)</p><p>The EB environment runs your app through a virtual environment name ‘venv’ on the EC2 instance, so we’ll need to manually install the missing packages there and not in the system’s environment.</p><p>Step 3: On the EC2 instance activate the virtual environment.</p><p>$ source /opt/python/run/venv/bin/activate</p><p>Step 4: Install the required packages in the virtual environment with sudo permissions and specify ‘pip’ in the virtual environment</p><p>$ sudo /opt/python/run/venv/bin/pip install ‘package_name’</p><p>Step 5: Deploy the app again and enjoy!</p><p>$ eb deploy</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ad5751859fb0" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>