 Hello my name is Marissa, I am an undergraduate at Cal Poly San Luis Obispo and at school I lead a team of students and together we write data driven stories. I also do software at Project Jupiter so I'm really happy to be here to share just a sliver of what I've learned with some of you about writing data driven stories. So foremost I want to assure you that you've seen many many data driven stories if you've used the web and so they all look very similar they look like this and so I want to first outline a few of the main elements that are shared across all of data driven stories and then I want to suggest to you a new type of element that you may or may not have seen and encourage you to use it if you ever decide to write a data driven story. So the three elements across any data driven story are charts, tables, and pros. So charts are just the graphs or the plots and these are summarizing the patterns or behaviors of your data and sometimes it compares different variables sometimes it compares different groups but overall it just summarizes general trends. If you want to show your readers a more specific value you can use a table and so that way your reader can pinpoint exactly the data point or the value you want to emphasize and then there's the part of a data driven story where the story comes in and that's the pros. So pros is just general plain language about what your data says and what that means for the reader and so if you're a reader that sees a lot of these you're very used to these types of elements but if you're a reader that is naturally curious about the world around them but maybe not so much about data in and of itself like some of us it might be overwhelming to see all of these elements at once and so whenever you're making a data driven story consider using one more type of element and that's the interactive component of it and so present your readers with some sort of interactive component that allows them to see a practical use case and play with some of the data so that they know or they're shown what you mean before you tell them and so I'm going to show you a few convincing examples that other people have done and use these type of interactivity in their stories and I'll walk you through why that's so important when you're thinking about the types of questions you want to address in your story and how you're going to convey your statistical methods to the readers so you may have seen the first one this is from the upshot it is called what is your twin city and you can go in and you can put in Portland as a city and it will spit out a city that is like Portland but not Portland so it has it has similar characteristics to Portland and here we see it's Kansas City, Missouri and you can also see that the writer has pulled apart different types of categories to see what other types of cities are like Portland but maybe voted differently or maybe is located on the east coast instead so as a reader if I'm going in and I'm changing input and output I'm able to compare these different values and I'm able to compare what the reader or what the writer is trying to tell me so through this type of interactivity readers can ask their own questions about how the data behaves and they can do their own comparative reasoning by changing the input and changing the output so I'm going to show you another example that's a bit more technical but still for a general audience and so this is called how algorithms know what you'll type next so anyone that uses their phone to text know that there are algorithms that exist where you can put in a word and it will predict what the next word you say will be and so this is trying to demystify how that works to a general audience and so without saying Markov chain or any type of complex statistical model or even a simple statistical model it shows you that if you go to change the word for different types of people the output will be different and readers know that the language the data set that is used to predict these values are coming from the Twitter accounts of the individual people so without knowing anything about probability as a reader I might guess that depending on the word I use in a in my Twitter account so if I use a specific word more maybe it'll predict that word more with higher probability and this is not exactly how it works and the reader might not know exactly how that works but the fact that they even have this preliminary idea of how something works is helpful for you as a writer because you can use their conception of how things work around them as a spring board to explain the actual statistical model that you used so you might say something like not only do people who analyze data care about how many times you say word in your language but maybe you care about the number of times you say that word following a specific word and so that way it's easy to understand and digest these concepts that are more technical but they already have some sort of intuition about the way things work around them the last example is a more practical example a little more scientific so I am from California and last year there was a lot of wildfires and so the New York Times decided to write an article with an interactive component about how the particulate matter from the smoke in the air is drifting from region to region and so this is and this is what it looks like now so it's not updated because all that has pretty much finished but you there was a lot of red in California and you can see some of that had drifted up to here Oregon and so because they put together this interactive component readers already see how it's related to them and they're more likely to stick with you through the rest of the article and see what you have to say about it so this type of interactive component gives readers sort of a stake in the end result they saw like how you want to present that information and they see it happening and now they want to see what you have to say as a writer about that end result I talked a bit about how my team at Cal Poly writes data-driven stories and I feel it would be very unfair not to show you some of those tools that we use so we use the Jupiter notebook which I'm sure you've heard about a lot this conference and so we write our code in the Jupiter notebooks and we can share it through GitHub so it makes it super easy to collaborate with different branches we also use Python and pandas so all of these libraries all work cohesively together so we chose Python and pandas as their data analysis libraries and it works together with plotly which is also a Python library and plotly is for graphing and the reason we specifically chose plotly is because there is a framework called dash and it allows you to create interactive web apps and so you can actually just copy the code from plotly and place them in dash and your plots will render so using these tools was super helpful and these are what we use to create our data-driven stories okay and also I would hate to leave you with the inability to go ahead and make your own so I actually put together a repo and you can fork the app.py file so it's a Python web app and if you fork it it gives you instructions in the readme on how to set up your own super simply and so you can get your own web app or data-driven story up and running. That's it, thanks!