In the comics, Pyle posts about ordinary human interactions conveying the humans as alien-esque creatures. I spent time analyzing all comics that have been posted to instagram in order to build a model that could predict the description of the post of any given comic. Nearly every post includes a one word description from the comic in the following format:. Over time, followers have begun guessing this word before they read it and announcing when they guess correctly.
This has led to discussions like:. This has all inspired me to see if I can create an algorithm that can guess for me. To more easily train my predictive model, I analyzed only comics that fit the following criteria:.
Almost every comic meets these criteria, giving me comics to work with. On average, Strange Planet comics have 35 words.
The shortest comic has 5 words and the longest comic has 74 words. The comics are also getting slightly longer as time goes on:. There does not appear to be a strong trend regarding how many words in to a comic the description appears.
The Strange Planet : And Other Stories
However, it is more likely that the description will show up in the third or fourth panel than in the first or second if the descriptive word is included more than once in a comic, all instances are shown below. I ran two models to look at what words were used most frequently in posts. I found that the top 20 most frequently used words in the comics fell in to the category of stop words, so I made a new model excluding stop words.
From the second word list, I also built a word cloud to show the most frequently used words in Strange Planet. From this analysis, I recorded the likelihood that a comic or description word has a positive sentiment. A few things to note from this sentiment analysis:. The description tends to be a longer word than the average word in the comics as is shown by the two histograms below. Thus far, a stop word has never been used as a description.
To both simplify my model and make it probably perform better, I assumed a stop word would never be used to describe a post. Of all the comics, only two descriptive words are used twice the rest are unique :. I made a similar assumption for the purpose of model building that a description word would never be resused. Even though I know that this is not true, it is the case frequently enough that I would prefer to use the assumption to be correct more often knowing that I will be incorrect when a word is duplicatted.
Typically, when you are creating a classification model, you want it to take input data and output a classification for that data:.
The Strange Planet : And Other Stories - sjewmisba.tk
Instead, we are looking at lists of varying lengths and predicting what word in that list is the description. Once I built the model, I processed my testing data in a slightly different way. For each word in the comic, I used the classification model to assign a probability of being the description to each word in the comic.
A separate model then selects the word with the highest probability as its guess for the description. Because of the method I am using to create my training data set, my data is split such that:. In order to address this, I used both the oversampling and undersampling techniques listed here on my training data. To convey how resampling changes the data my model is built on, here are two pair plots showing some of the input features my model is using. In the plots below, blue dots signify words that are in the comic but not the description. Orange dots signify description words. To the left is the data before I rebalanced.
The right is the data after I rebalanced it. Notice how much larger a percent of the points are orange in the right image. Notice also that some of the observations listed in the EDA section can be seen in the scatterplots and histograms above. For example:. I am sad to say that my model performs pitifully. This series will celebrate the natural wonders that remain and reveal what we must preserve to ensure that people and nature thrive. The message is clear. We can still fix it. Our Planet is a eulogy, a confession, a slap on the wrist, a call to arms.
Read: What scientists learned from strapping a camera to a polar bear. There is optimism, too.
airtec.gr/images/app/2724-como-hacer-para.php Amid doom-laden warnings, the documentary highlights success stories in which conservation measures have allowed species to start bouncing back. When we watch five cheetah siblings do their best lion impressions and cooperatively bring down a wildebeest, Attenborough tells us that we get to enjoy such dramas only because the Serengeti has been protected for decades.
And in a sequence of unexpected poignancy, wild horses, foxes, and wolves are seen thriving among the ruins of Chernobyl, the radiation a minor inconvenience compared with the boon of human absence. Some episodes still feel as disjointed as those of Planet Earth II did, with few narrative threads connecting the individual sequences.
And the score never goes for a subtle musical cue when a saccharine one will do. But these are small gripes for a series that audaciously treads where its predecessors have feared, and sets the bar for its successors. Even the BBC are now saying that they want environmental messaging in their programs.
And not a moment too soon. We want to hear what you think about this article. Submit a letter to the editor or write to letters theatlantic. A dust storm blows in over a colony of Socotra cormorants. Ben Macdonald. That remain! Ed Yong is a staff writer at The Atlantic , where he covers science.
Related The Strange Planet : and Other Stories
Copyright 2019 - All Right Reserved