<?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 Alexei Scerbakov on Medium]]></title>
        <description><![CDATA[Stories by Alexei Scerbakov on Medium]]></description>
        <link>https://medium.com/@alexei.scerbakov?source=rss-1c597d412914------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Fg8mPzV4dnWbOZMFeyCqjA.jpeg</url>
            <title>Stories by Alexei Scerbakov on Medium</title>
            <link>https://medium.com/@alexei.scerbakov?source=rss-1c597d412914------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 04:04:42 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@alexei.scerbakov/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[Facility price prediction with Azure ML]]></title>
            <link>https://medium.com/@alexei.scerbakov/facility-price-prediction-with-azure-ml-aa7dae17b8f1?source=rss-1c597d412914------2</link>
            <guid isPermaLink="false">https://medium.com/p/aa7dae17b8f1</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-functions]]></category>
            <category><![CDATA[microsoft]]></category>
            <category><![CDATA[azure-machine-learning]]></category>
            <dc:creator><![CDATA[Alexei Scerbakov]]></dc:creator>
            <pubDate>Mon, 28 Jan 2019 08:18:26 GMT</pubDate>
            <atom:updated>2019-01-28T08:18:26.672Z</atom:updated>
            <content:encoded><![CDATA[<p>Out of personal interest for the real estate market in my home town, Graz, Austria I decided to make use of machine learning to help me get a good feel for the market and in the process explore modern technologies like Azure Functions and Azure Machine Learning. My idea was to gather housing data and analyze it via Azure machine learning to be able to predict prices for new objects.</p><h3>The parser:</h3><p>The bread and butter of every good machine learning model is the data provided, which meant that I needed to collect it! I ended up building a parser in .NET using the HtmlAgilityPack to parse object data from the biggest Austrian website for real estate. I am currently collecting data of one city, but the parser can be adjusted to collect data from different sources (sites) and of different cities. Even when collecting data of only one city only I ended up with ~2000 datasets + ~200 datasets/week.</p><p>To collect the data periodically a scheduled Azure Function was used, which runs once a week and saves the data in an Azure Table. Azure Functions in combination with an Azure Table are particularly suited for these tasks. Azure makes hosting this scheduled code painless, and the Azure Table keeps the “database” management part to a minimum. It is easy to export the data in a .csv format which is then used in the Azure ML studio.</p><h3>Machine learning:</h3><p>For the machine learning part Azure machine learning studio was used, which is a very simple way to make use of a machine learning algorithm to make predictions. All popular machine learning algorithms are available, and in my case “Bayesian Linear Regression” was used, to train my prediction model. Azure ML offers a simple interface to build a training model and create a web API endpoint to be able to make predictions for new data.</p><p>This web API will again interact with an Azure Function (but this time an HTTP triggered one. The flow then looks like this: the client side code calls the Azure Function, which then calls the web API endpoint and returns the predicted data.</p><h3>Client side form:</h3><p>For the client side I am using React with Office UI fabric. React just for learning purposes and Office UI fabric to easily make it look good.</p><p>The overall architecture overview would be as follows:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d0cIrNNKbNHH_LYrLSZUnw.png" /></figure><p>You can give the whole experiment a try at this <a href="http://scerbakov.com/price-prediction/">URL</a>.</p><p>As mentioned the prediction model is limited to the city of Graz, Austria!</p><p><strong>Thank you for reading and have fun testing the app/program. Any kind of feedback would be greatly appreciated!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aa7dae17b8f1" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>