<?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 OzzyCodes on Medium]]></title>
        <description><![CDATA[Stories by OzzyCodes on Medium]]></description>
        <link>https://medium.com/@ozzycodes?source=rss-336de5a3d545------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*qykMaI2hp5RU806uuP2-Ew.png</url>
            <title>Stories by OzzyCodes on Medium</title>
            <link>https://medium.com/@ozzycodes?source=rss-336de5a3d545------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 17:34:08 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ozzycodes/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[What Happens When You Make Three AIs Fact-Check Each Other]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://pub.towardsai.net/what-happens-when-you-make-three-ais-fact-check-each-other-03b3d04f7a73?source=rss-336de5a3d545------2"><img src="https://cdn-images-1.medium.com/max/2600/0*xG50LIsGfB7BTb5y" width="3840"></a></p><p class="medium-feed-snippet">A reproducible method for measuring AI calibration &#x2014; and an unexpected lesson from building it.</p><p class="medium-feed-link"><a href="https://pub.towardsai.net/what-happens-when-you-make-three-ais-fact-check-each-other-03b3d04f7a73?source=rss-336de5a3d545------2">Continue reading on Towards AI »</a></p></div>]]></description>
            <link>https://pub.towardsai.net/what-happens-when-you-make-three-ais-fact-check-each-other-03b3d04f7a73?source=rss-336de5a3d545------2</link>
            <guid isPermaLink="false">https://medium.com/p/03b3d04f7a73</guid>
            <category><![CDATA[llm-evaluation]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[claude-code]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[gemini]]></category>
            <dc:creator><![CDATA[OzzyCodes]]></dc:creator>
            <pubDate>Sat, 20 Jun 2026 16:13:10 GMT</pubDate>
            <atom:updated>2026-06-29T04:26:21.244Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Simple Passive Incomes]]></title>
            <link>https://medium.com/@ozzycodes/simple-passive-incomes-be3d958a3e46?source=rss-336de5a3d545------2</link>
            <guid isPermaLink="false">https://medium.com/p/be3d958a3e46</guid>
            <category><![CDATA[selling]]></category>
            <category><![CDATA[non-technical]]></category>
            <category><![CDATA[side-hustle]]></category>
            <category><![CDATA[passive-income]]></category>
            <dc:creator><![CDATA[OzzyCodes]]></dc:creator>
            <pubDate>Sat, 11 Jun 2022 03:57:38 GMT</pubDate>
            <atom:updated>2022-06-11T03:57:38.771Z</atom:updated>
            <content:encoded><![CDATA[<p>Over the past few months, I have been looking for simple ideas for adding income streams that don’t take up much time.</p><p>I’ve read many blogs in regards to creating bots online, investment ideas (way outside my budget for a steady income stream), to simple applications.</p><h4>A Few Simple Ideas to get Started…</h4><p>As many of you might have heard of, sharing unused internet bandwidth is an easy way to generate a passive income. Apps like <a href="https://r.honeygain.me/OZZCO00382"><strong>Honeygain</strong></a><strong> </strong>which can be connected to JumpTask which provides 50% bonuses on all rewards earns will make you some money, but won’t make you rich.</p><p>I’ve been using Honeygain’s app, utilizing the JumpTask wallet, which I then connect to my MetaMask wallet for transferring funds. My overall opinion of this automated passive income stream is; yes it works. I use my computer everyday as I work mostly remote as a Data Engineer. The running app does not affect my PC performance when connected to cloud services, so I don’t mind having it run in the background at all times. I also use my PC as my main gaming console, and have not seen any lag in game play due to the Honeygian app running in the background.</p><p>Another simple income stream is <a href="https://surveys.gobranded.com/users/register/f279f8c0247e66c7068fbc000"><strong>Branded Surveys</strong></a>. Branded Surveys is a survey site (as well as a mobile app) to gain points for each survey completed. The more survey’s you complete, the more rewards you can obtain through their Branded Elite program. The program is based on the number of surveys you take per month.</p><p>Branded Elite badge unlocks access to bonus points for your weekly survey completions. Remember, payouts for <strong>Silver &amp; Gold Badge</strong> members are instantly approved! A brief overview of their Elite program:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1012/1*yPSZJ3slQ_PADdALEjzS4g.png" /></figure><p><strong>The best part of branded elite is that you can redeem your point to your PayPal account for instant cash transfers!</strong></p><p>One more option, which I’ve found to be quite profitable is selling used books. Used books are typically tossed out and given away to local thrift shops which you can buy for cheap and possible make a decent profit, <strong>$15–$50 per book</strong> from my experience. I mostly use eBay as a selling stream for books, just do a bit of research on your available book source and find it’s worth!</p><h4>Give a couple of these options a try and see if they are right for you to make a decent passive income!</h4><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=be3d958a3e46" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Analyzing Location data for suggesting a Bar Location — Data  Science with Python]]></title>
            <link>https://medium.com/@ozzycodes/analyzing-location-data-for-suggesting-a-bar-location-data-science-with-python-760ea871f78f?source=rss-336de5a3d545------2</link>
            <guid isPermaLink="false">https://medium.com/p/760ea871f78f</guid>
            <category><![CDATA[pandas]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[k-means-clustering]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[foursquare-api]]></category>
            <dc:creator><![CDATA[OzzyCodes]]></dc:creator>
            <pubDate>Sun, 26 Jul 2020 17:18:34 GMT</pubDate>
            <atom:updated>2020-07-26T17:18:34.632Z</atom:updated>
            <content:encoded><![CDATA[<h3>Analyzing Location data for suggesting a Bar Location — Data Science with Python</h3><h4>A final capstone project for a 9 course series, Data Science Certification; <a href="https://www.coursera.org/learn/applied-data-science-capstone/home/welcome"><em>IBM Data Science Professional Certificate</em> </a>via <a href="https://www.coursera.org/">Coursera</a></h4><p><em>Utilizing Python and its abundance of Data Science related libraries, we will resolve the business problem below.</em></p><p><strong>Python tools/ libraries used:</strong></p><ul><li>Folium library, including choropleth map, heatmap in map view</li><li>Honeycomb Grid for Folium map</li><li>Google Geocoding APIs</li><li>Foursquare APIs</li><li>K-Means Clustering Algorithm</li><li>Bar Charts</li><li>Pandas, Numpy, Shapely</li></ul><h4><strong>The Business Problem — Description:</strong></h4><p>For this project, we are going to utilize the Foursquare API for referencing <strong>location data in the Boston, MA</strong> area. This data will allow us to report on and provide feedback to interested parties <strong>looking at locations in order to open a bar in an neighborhood</strong>.</p><p>Boston, MA is a very popular city when it comes to nightlife and entertainment as well as its ratings for recommended bars. With this in mind, we would want to <strong>identify all of the locations that are not overly crowded</strong>, in terms of the <em>number of bars </em>(this may setup a new hot-spot location for more restaurants to follow). BUT, we want to remain in a popular tourist area as to not stray from highly recommended areas.</p><p>With the use of Python and its vast amount of Data Science libraries, we can accurately predict where to set-up shop for a new bar (or at least recommend).</p><h4>The Data</h4><p>Following the Business Problem, the data necessary to achieve accuracy will need to account for situations and factors such as:</p><ul><li>Demographic information, e.g. population, density, education, age, income</li><li>Number of tourist attractions in the neighborhood and/or nearby</li><li>Number of existing restaurants in the neighborhood and/or nearby</li><li>Total traffic in the different neighborhoods (more populated the better)</li></ul><p>We are going to stay within a certain radius of the city center, for defining our neighborhoods in a grid-like fashion.</p><ul><li>Centers of potential locations will be generated with the use of algorithms and approximate addresses of the potential centers</li><li>Center locations will be obtained using <strong>Google Maps API reverse geocoding, and the use of Google Maps API geocoding </strong>for obtaining location coordinates.</li><li>The number of bars and their type/location in every neighborhood will be obtained using <strong>Foursquare API</strong></li></ul><h4>Getting Started</h4><p>Import the Data</p><pre>import wget<br><br># GeoJSON Data<br>wget.download(<br>    &#39;http://bostonopendata   boston.opendata.arcgis.com/datasets/3525b0ee6e6b427f9aab5d0a1d0a1a28_0.geojson?outSR={%22latestWkid%22:2249,%22wkid%22:102686}&#39;,<br>    &#39;boston_data.json&#39;<br>)</pre><h4>A Quick look at the map of the GeoJSON datafile on a map using folium</h4><pre>boston_map = folium.Map(location=[42.329143, -71.076449], zoom_start=11)<br>boston_map.choropleth(<br>    geo_data=boston_data<br>)<br><br># display map<br>boston_map</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*Ebu6qX28uZSJuaLt0pXFMw.png" /></figure><p>Looking at the raw data, the <strong>features</strong> key holds the relevant neighborhood data, from the GeoJSON file. Transform this into a dataframe.</p><pre>boston_neighborhoods = boston_data[&#39;features&#39;]</pre><pre># define the dataframe <br>columns column_names = [&#39;Neighborhood&#39;, &#39;SqMiles&#39;,&#39;Latitude&#39;, &#39;Longitude&#39;]  </pre><pre># instantiate the dataframe <br>neighborhoods = pd.DataFrame(columns=column_names)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/863/1*uKxgZgxdGcnUGp-A76baWw.png" /><figcaption>Initial dataframe format from provided dataset</figcaption></figure><p>From the provided dataset, there is some work to be done to clean the Latitude and Longitude columns. The data produced was in a MultiPolygon geometry feature and we would want to convert the coordinates to a single Point. In this example I used the <strong>jmespath python library:</strong></p><pre><em># Query the results you want from the json dataset,<br># and in this case we have neighborhoods with different size arrays #so multiple queries are needed</em><br><br>query = &quot;&quot;&quot;<br>[*].{<br>Neighborhood: properties.Name,<br>SqMiles: properties.SqMiles,<br>Latitude: geometry.coordinates[0][0][0][1],<br>Longitude: geometry.coordinates[0][0][0][0]<br>}<br>&quot;&quot;&quot;<br><br>query2 = &quot;&quot;&quot;<br>[*].{<br>Neighborhood: properties.Name,<br>SqMiles: properties.SqMiles,<br>Latitude: geometry.coordinates[0][0][1],<br>Longitude: geometry.coordinates[0][0][0]<br>}<br>&quot;&quot;&quot;<br><br>compiled = jmespath.compile(query)<br>results = compiled.search(boston_neighborhoods)<br>new_neighborhoods_df = pd.DataFrame.from_records(results)<br><br>compiled2 = jmespath.compile(query2)<br>results2 = compiled2.search(boston_neighborhoods)<br>new_neighborhoods_df2 = pd.DataFrame.from_records(results2)<br><br><em># Merge the two dataframes created from the above queries</em><br>merged_neighborhood_df = pd.merge(new_neighborhoods_df, new_neighborhoods_df2, on=&#39;Neighborhood&#39;)<br><br><em># Using the Mask function, we can drop all the null values of the <br># merged dataframes and combine all data<br># columns by the specific column name</em><br>merged_neighborhood_df[&#39;Latitude&#39;] = merged_neighborhood_df[&#39;Latitude_x&#39;].mask(<br>    pd.isnull, merged_neighborhood_df[&#39;Latitude_y&#39;])<br>merged_neighborhood_df[&#39;Longitude&#39;] = merged_neighborhood_df[&#39;Longitude_x&#39;].mask(<br>    pd.isnull, merged_neighborhood_df[&#39;Longitude_y&#39;])<br>merged_neighborhood_df[&#39;SqMiles&#39;] = merged_neighborhood_df[&#39;SqMiles_x&#39;].mask(<br>    pd.isnull, merged_neighborhood_df[&#39;SqMiles_y&#39;])<br><br><em># create a final dataframe with the previously given column names,  # for a clean dataset</em></pre><pre>final_neighborhood_df = merged_neighborhood_df[column_names]<br>final_neighborhood_df.head(10)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/426/1*6N9RDr2e96oCZpZAnYMW8w.png" /><figcaption>LOOKS MUCH BETTER!</figcaption></figure><p>This is just ten of the 26 neighborhoods in Boston used for the analysis. Now I would like to superimpose all of the Boston Neighborhoods on top of the Boston map previously created with folium (these location markers will be created from the coordinates for each neighborhood).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/867/1*ZkyTnycak0joyXhvGzZiLg.png" /></figure><h3>Foursquare Analytics</h3><p>Next, we are going to start utilizing the Foursquare API to explore the neighborhoods and segment them.</p><h4>Define Foursquare Credentials and Version</h4><pre>CLIENT_ID = &#39;your_id&#39; # your Foursquare ID<br>CLIENT_SECRET = &#39;your_secret&#39; # your Foursquare Secret<br>VERSION = &#39;current_version&#39;</pre><h4>Let’s get the top 100, most popular venues that are in Boston within a .5 mile radius (804 meters).</h4><p>After obtaining the most popular venues, next will be to places all the venues into a dataframe, assigning them to each neighborhood.</p><h4>Next we want to filter for Bars/ pubs only for the purpose of this analysis:</h4><pre>boston_bar_vens = boston_venues<em>[</em>boston_venues<em>[</em>&#39;Venue Category&#39;<em>]</em>.str.contains<em>(</em>&#39;Bar|Pub&#39;<em>)]</em>.reset_index<em>(</em>drop=True<em>)</em></pre><pre>boston_bar_venues = boston_bar_vens<em>[</em>~boston_bar_vens<em>[</em>&#39;Venue    Category&#39;<em>]</em>.str.contains<em>(</em>&#39;Barbershop&#39;<em>)]</em>.reset_index<em>(</em>drop=True<em>)</em></pre><pre>boston_bar_venues.index = np.arange<em>(</em>1, len<em>(</em>boston_bar_venues<em>) </em>+ 1<em>)</em></pre><pre>boston_bar_venues<em>[</em>&#39;Venue Category&#39;<em>]</em>.value_counts<em>()</em></pre><p>The resulting dataframe; Venue Name and Category allocated to each neighborhood:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/861/1*IauWuCrZ98_h82TSTworPQ.png" /></figure><p>After allocating all of the correct venue types to our new dataframe, we can begin digging into the data.</p><p>First I started with a simple count of total bar venues (not the total number of bars) in all of the neighborhoods in Boston and created a visual for easy understanding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/373/1*kcOfusyu60c8ed61_Xz3Ww.png" /></figure><p>These venue types came from a total of 12 unique venue types that are within the Bar/ Pub parameters.</p><h3>Analyzing Each Neighborhood</h3><p><strong>What are the most common venue categories in Boston’s neighborhoods?</strong></p><p>Below you can visually see the frequency of the type of bar and pub venue pops ups more often than others:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/852/1*RPcRrCXe5wfNosZReNa9sw.png" /></figure><p><strong>Next, let’s group rows by neighborhood and taking the mean of the frequency of occurrence of each category</strong></p><pre># one hot encoding<br>boston_onehot = pd.get_dummies(boston_bar_venues[[&#39;Venue Category&#39;]], prefix=&quot;&quot;, prefix_sep=&quot;&quot;)<br><br># add neighborhood column back to dataframe<br>boston_onehot[&#39;Neighborhood&#39;] = boston_bar_venues[&#39;Neighborhood&#39;]<br><br># move neighborhood column to the first column<br>fixed_columns = [boston_onehot.columns[-1]] + list(boston_onehot.columns[:-1])<br>boston_onehot = boston_onehot[fixed_columns]<br><br>boston_onehot.head()</pre><p>This gives you the ability to obtain the mean frequency for each venue by Neighborhood, and placing the values into a dataframe for further analysis.</p><p>After creating the above dataframe, we have the ability to display the top 10 bar and pub venues by neighborhood in order:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2gzHZb9JfgpR0M4IowIaIg.png" /></figure><p>The next task is to summarize these neighborhoods and venues by a means of clustering.</p><h3>Cluster Neighborhoods</h3><p>With the commonality between venue categories in each of the neighborhood, it helps to use the unsupervised learning <strong>K-means algorithm </strong>to cluster the neighborhoods. K-Means algorithm is one of the most common cluster method of unsupervised learning.</p><p>First, we would want to obtain the optimal number of clusters to consider for K-means clustering. I started off by creating a graph using matplotlib and the sklearn Silhouette Score libraries.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/954/1*QqeNXS8Ce3Axq6oMMWB-Bg.png" /><figcaption>Optimal Cluster Graph</figcaption></figure><p>From the graph you can determine that 8 is the optimal number of clusters to be used. Below if the merged table with cluster labels for each of the neighborhoods.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vikX0wUHupcFDMxGeU6u5Q.png" /><figcaption>Displaying the first 5 neighborhoods in the dataframe</figcaption></figure><p>We will also have the ability to estimate the number of 1st Most Common Venue’s in each cluster. Using a bar-chart would be a good visualization to do-so.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/408/1*AgvaCrQ6l4TPyfQWzW8n8A.png" /></figure><p>Finally, we can then visualize the different clusters created on a map of Boston, to get an idea of where each venue category is located.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/389/1*-HMVvuoIbnw1TpNJzHNJHQ.png" /></figure><h3>Results</h3><p>For my analysis of the best locations for opening a new restaurant in Boston I created a datatable for each cluster to get an idea of what types of bars were ranked from 1st Most Common to 10th Most Common.</p><p>Looking at the Cluster Map with each neighborhood, it can be determined that the most popular bar areas are between South Boston to the South Boston Waterfront and up to Downtown Boston. The Overall most popular bar types are a typical bar along with a Pub (Irish Pub’s makes sense as this is Boston!).</p><p>Below is an overall summary I had created to organize each cluster with the ranked bar venue’s:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*XzWyrBMDB4FnFzh2A0Xnfw.png" /></figure><h3>Discussion</h3><p>Boston has a large number of nightlife venues, aside from the focus of this analysis, Bars and Pubs. With a total of 26 neighborhoods, the most popular of bar type can vary depending on the area. Due to the complexity of of variables for this analysis many different clustering approaches and classifications can be used. Although, using a classification method as described, a specified targeted venue type can return high quality results.</p><p>This study could be further analyzed using more data, which could include neighborhood income and population along with crime rates and cost of living factors.</p><p>My study was concluded through the use of clustering the specified venue category by neighborhood and displaying it as a visual on a map of Boston.</p><h3>Conclusion</h3><p>With the availability of open resources to dataset like the one I had used for obtaining the <a href="http://bostonopendata-boston.opendata.arcgis.com/datasets/3525b0ee6e6b427f9aab5d0a1d0a1a28_0.geojson?outSR={%22latestWkid%22:2249,%22wkid%22:102686}">Boston data</a>, a much more complex analysis could be used. But taking narrow measures as I had for a small category of venue’s can create a quality idea of business venture opportunities.</p><p>Data is key to all industries and with the ability to obtain and analyze it, there is no limit to what can be achieved.</p><blockquote>Jupyter Notebook Complete Code:<br><a href="https://view.datalore.io/notebook/ZCtVlAiDKkrK46ofOHAYcS">https://view.datalore.io/notebook/ZCtVlAiDKkrK46ofOHAYcS</a></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=760ea871f78f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>