Halloween is rapidly approaching, and panic is setting in. What are you going to go as this year? Will it be the scary zombie, the topical GoT character, the fun filled group costume or something that shows off a little leg? This weeks webinar is for those of you who haven’t quite decided what to be for Halloween yet. I’ve got a very special (return) guest with me, the wonderful and talented Jewel Loree from Tableau, who is going to help me build a Halloween costume dashboard. As you can see, Jewel and I already have our costumes sorted; but we know how hard it can be to decide. Which is why we’ve turned to the trusty data to help you figure out whether to go scary or sexy this year!
Boo! Data! Ah!
We used a spooky Crawler to get all the costume data from Buy Costumes. For this Crawler we drilled down to the individual costume pages to get the maximum amount of data for each costume. Because we’re on an individual costume’s page, we used a single results extraction.
For each costume we extracted the name, image, price, average reviews, number of reviews and description. I mapped pretty much all these fields as text because it’s easier for Jewel to put into Tableau that way, but of course you guys know that I could use our column types to map the data as numbers, images, currency, etc.
Once you’ve mapped you data, you need to train at least 5 pages. Luckily you shouldn’t need to do any more training because the tool will remember the training you did on the first page. What you’re looking for is to make sure that the app has recognized the right data. If it hasn’t you can do some extra training to help it recognize the data pattern better.
Once we’ve finished mapping our data, its time to run the crawler. To help Jewel out a little bit I actually ran a separate crawler for each of the different costume categories – so that when we build the dashboard we can filter by category. Finally, all I have to do is download all the Crawled data as an Excel file and slide it across the table to Jewel so she can do her data vizzing thang!
Oooo. Hello there data viz
Once you’ve got your Excel file you’re going to need to do a little bit of data cleansing before you can upload it to Tableau. The first thing you need to do is make sure your column headers are in the first row of your Excel table. Then, delete any columns you don’t need.
One of things Jewel wanted to make was a word cloud, so to do that she did a little excel magic. Move the product ID column and the costume description column to a new sheet and use the “Text to Columns” feature to move each word to a new column. Then use the reshape data tool to turn each of those columns into one row. Because the product ID is the same, when you bring it into Tableau you can match the words to the costume.
Then Jewel uploaded both Excel files into Tableau – one with the words and one with the data from import. Then she created three charts with the data by dragging and dropping each of the columns she wanted to map into the table.
If you’re not sure what to make you can use the “show me” function which will suggest different types of visualizations you can make based on the data columns you have selected. Once you’ve made all you vizzes you can bring them into a dashboard, by simply dragging your charts onto it.
For more detailed information on how Jewel made her charts, check out her blog post.
For more on how to use Tableau Public, check out their tutorials.
What if the data is not on the screen but maybe behind a drop down menu?
You can use an XPath to get this data! I even did a whole webinar on it.
Does it matter if the data is in slightly different places on each subsequent page?
It depends on how different of a place the data is in. If it is within the same XPath you can just do a little extra training (like i did in the webinar) to expand the results slightly to accommodate this. If it moves around a lot, you may need to write some custom XPath to get to it.
Can import.io be used to extract images from a site?
We sure can! Just change the column type to image.
Can import.io handle wikipedia tables?
We can indeed! There’s even have a feature (Auto Table Extract) that allows you to extract tables in one click!
What’s the airspeed velocity of an unladen swallow?
I’m afraid I can’t answer this without more information…do mean African or European?
Join us next time
We’ve been doing a lot of joint webinars recently, which has been great fun, but next week we’re going to take it back to basics. Bamford and I will be reunited to give you the lowdown on all the different ways you can use import.io to extract data from the web. This is also your chance to ask us anything you like – and we mean anything – so come prepared with questions!