<?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 Grant on Medium]]></title>
        <description><![CDATA[Stories by Grant on Medium]]></description>
        <link>https://medium.com/@grantivisonlane?source=rss-2faca50cca74------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*UfhVeUzJZeBXzYX0.jpg</url>
            <title>Stories by Grant on Medium</title>
            <link>https://medium.com/@grantivisonlane?source=rss-2faca50cca74------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 02:14:23 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@grantivisonlane/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[Measuring Outrage — Bears Ears NM]]></title>
            <link>https://medium.com/@grantivisonlane/measuring-outrage-bears-ears-nm-3e8607303191?source=rss-2faca50cca74------2</link>
            <guid isPermaLink="false">https://medium.com/p/3e8607303191</guid>
            <category><![CDATA[environment]]></category>
            <dc:creator><![CDATA[Grant]]></dc:creator>
            <pubDate>Sun, 17 Feb 2019 01:42:07 GMT</pubDate>
            <atom:updated>2019-02-17T01:42:07.275Z</atom:updated>
            <content:encoded><![CDATA[<h3>Measuring Outrage — Bears Ears NM</h3><p>When it comes to our opinions on public lands such as our parks, monuments, and recreation assets that we all share, it’s important for our voices to be heard.</p><p>With this in mind, I aggregated social media data to track public opinion. By mining data on Twitter, I collected a sample of over 17,000 user tweets related to President Trumps Executive Order aimed at removing protections from Bears Ears National Monument.</p><h3>1. Download Tweets</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/0*k9ULkqvrKdI4LVHy." /></figure><p>By using the search function of the Twitter RestAPI I collected the following amount of tweets containing the phrases below:</p><p>“#BearsEars” — 5704 Tweets</p><p>“Bears Ears” — 12080 Tweets</p><p>These tweets range from 4/20/2017–4/26/2017, which is the max date range that Twitter allows for searching (about 6 days back)</p><h3>2. Remove Duplicates</h3><p>To account for the people who tweeted “#BearsEars” and “Bears Ears” in the same tweet, I used the following logic to remove duplicates.</p><p><strong>Tweets = (Bears Ears — #BearsEars) + (#BearsEars — Bears Ears)</strong></p><h3><strong>3. Tweet and Word Distributions</strong></h3><p>This leaves us with the following tweet distribution</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1022/1*-b1WKX59w8tJ5C26p1NrZg.png" /></figure><p>From the data, we can see that during this time period people re-tweeted <a href="https://medium.com/u/1756c73a8e19">Jon Favreau</a>, National Geographic, Senator Tom Udall, AltSmithsonian and the organization Save Bears Ears the most.</p><p>Lets look at the word distribution:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/0*OsBI0H2f4gJTqawE." /></figure><p>It’s clear retweets are heavily influencing the data. Lets look into what people are actually typing on their keyboards by removing retweets.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*iHqi_nnafzYlR-LWp_QTOQ.png" /></figure><p>without retweets we have over 1700 completely orginal tweets about bearears from twitter users</p><p>Lets look at a word map of the data:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/533/1*IkYirajaa1pwKsnjpoFg7w.png" /><figcaption>Unformatted tweet data in a word cloud</figcaption></figure><p>we can see some sentiment trends from this information, lets try and polish it up using only nowns, adjectives, and verbs</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/448/1*opQvM2QGNBFv8-WQQVoKYQ.png" /><figcaption>Formatted tweet data in a word cloud</figcaption></figure><p>Interesting, but we can take it a step further to understand the sentiment of tweeters using sentiment analysis.</p><h3>3. Sentiment Analysis</h3><p>Sentiment analysis classifies a tweet as positive, neutral, or negative. It does it though <strong>word scoring</strong>, which is the process of assigning a score to each word in a tweet and summing it up to get a sense of the general attitude.</p><p>I ran the 1700 original Bears Ears NM related tweets through 2 sentiment analysis programs: Sentiwordnet’s “VADER”, and TextBlob. For comparison, I took a sample of 200 tweets and scored them myself to try and get a gage of how accurate these tools are for this specific political topic. Lets take a look at how these tweets scored.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/427/1*F86Q4l6zJoLHRu37giV44A.png" /><figcaption>Results of Sentiment Analysis using Python</figcaption></figure><p>Overall we see a negative trend developing related to this executive order, with about 4/10 tweets being explicit in frustration, and only 1/10 supporting it from my own actual analysis (going through each tweet and assigning it a score).</p><p>We can see that TextBlob is not close to reality (actual analysis), this makes sense when we consider the program is calibrated to understand the tone of movie reviews, and likely does not understand the nuances of environmental politics. Vader does significantly better with respect to negative tweets, likely because it was created to analyze topics closer to this one, but is still is a bit off. Lets try and understand why.</p><p>VADER is looking through each tweet and scoring words such as “bad” or “terrible” and calls those tweets negative, while words like “good”, “nice”, and “excellent” are scored as positive. While VADER understands more slang terms unlike Textblob, the issue isn’t that it doesn’t have the vocabulary, but rather that someone could be using angry words in support of the presidents executive order, and nice words opposing it! For example Vader scored this tweet:</p><p>“Respect the Antiquities Act don’t rescind the National Monuments designation save BearsEars”</p><p>as positive, when in fact it’s very negative when respect to the context of this argument (anti-executive order).</p><p>Ultimately, this example analysis goes to show that the machine learning program used for any sentiment analysis needs to be calibrated to <em>that exact </em>political discussion to understand the context<em>.</em></p><p>This is just an introductory analysis of this topic using Twitter, but it’s clear there is some value to this data (once in the right technical calibration). The next step would be other twitter based analysis methods , as well as using geography, other social media sites to further complete the dataset.</p><p>Thanks for reading!</p><p>Grant</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e8607303191" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>